Skip to main content

Full text of "Climate Throuch The Ages"

See other formats



m< OU_1 60521 >m 

^ co 


By the same author 







I.S.O., D.Sc., F.R.Met.Soc. 


First published 1926 
Revised edition 1949 
Second impression 1950 




SOME years ago I had the privilege of laying before the 
public " The Evolution of Climate." As its title 
implies, that book mainly consists of an account of the 
variations of climate during geological times, and it deals 
especially with the Quaternary Ice-Age and the Post-glacial 
period. The causes of the events described were touched on 
only very briefly, and chiefly for the purpose of emphasising 
the climatic importance of the period of continental emergence 
which intervened between the mild -climates of the Mesozoic 
and Tertiary periods and the beginnings of the Ice-Age. It 
was my intention then to write a companion volume dealing 
with the mechanism of climatic changes and drawing on the 
first volume for illustrations. But there were many problems 
requiring further investigation and much reading to be done 
before such a work could be approached with any confidence, 
and so for several years desire outran performance. Then 
came the work of Wegener on the theory of continental drift, 
and of Koppen and Wegener on the interpretation of the 
climatic record in terms of the travels of the continents across 
the parallels of latitude, and this stimulated me to complete 
the investigation in order to examine Wegener's theory from 
the climatic side. 

" Climate through the Ages " has not been altogether easy 
writing ; the present is the key to the past, but there have 
been many strange meteorological situations for which we 
have no present parallel to guide us. Moreover, the 
meteorology of the present has still to solve many problems 
of its own, and I am even encouraged to hope that the 
meteorology of the past may at times help in the study of the 
present. The theory of the circulation of the earth's 
atmosphere, for instance, is not yet complete, and it may be 
that the modifications of the circulation during the varied 
climatic history of the globe, as deduced from the distribution 
of rainfall and temperature, will provide just the additional 
material required for a solution.* 


The point of view being meteorological, it has not been 
considered necessary to keep to a strict chronological order 
in discussing the climates of different geological periods, as 
was done in " The Evolution of Climate." For convenience* 
of reference, a list of the geological periods, with some idea 
of their duration and a brief note on the general type of 
climate, has been given in Appendix I. For the purposes of 
the discussion, geological climates are classed as " warm " 
or " glacial." We are living now in a " glacial " period, 
though fortunately not at its maximum, and the meteorological 
conditions during the Quaternary Ice-Age are not really 
strange to us. The " warm " periods, in which a genial 
climate extended almost or quite to the poles, are meteoro- 
logically much more remote. In the best developed warm 
periods it is probable that ice was unknown. This approach 
to uniformity of temperature was inevitably associated with 
very great changes in the winds, the ocean currents, and the 
distribution of rainfall. The way in which this situation came 
about was not simple, and I have thought it best to open the 
book by setting out, in an Introduction, the geological evidence 
as to the existence and nature of these warm periods, and 
especially the extent to which climatic zones were developed. 
This greatly simplifies the subsequent meteorological discussion 
by allowing us to take the main facts for granted and to 
consider the variation of the individual factors of climate one 
by one, until the ground has been prepared for a more complete 
and logical discussion in Chapter X. 

" Warm " periods have been the rule in the history of 
the earth, but from time to time the revolutionary forces 
of the underworld have succeeded in overthrowing this 
quiet and genial existence and have brought about crises in 
the earth's history, and the greatest of these crises have been 
marked by ice-ages. There have been at least four major 
ice-ages, in the early Proterozoic, in the late Proterozoic or 
Algonkian, in the Upper Carboniferous, and in the Pleistoce;ne- 
Recent periods. Of the first two of these we know little, but 
that little suggests that they were entirely analogous to the 
Quaternary. The Upper Carboniferous glaciation, on the 
other hand, was highly abnormal, in that the greatest ice- 
sheets developed in regions which are now not far from the 
equator. It seems probable that there was another great 


period of mountain-building at the close of the Cretaceous, 
which developed in such a way that the distribution of land 
and aea, and of mountain ranges, was not favourable for 
extensive glaciation. 

The book* is divided into three sections. In the first, the 
various factors of climate are discussed, and the scope which 
they offer for the introduction of climatic changes is considered, 
quantitatively when possible. This part of the work is 
essentially a text-book of meteorology, in which, however, 
some of the constants of the ordinary meteorological text-book 
ar.e treated as variables. Various theories of climatic change 
are discussed in successive chapters as they arise, but no 
attempt has been made to include all the theories which have 
been put forward from time to time. No useful purpose 
would be served by solemnly refuting, for instance, the theory 
that the mild polar climates of the Tertiary were due to 
radiation from a warm moon, while the author of another 
theory of the warm periods, who attributes them to the heat 
set free by decomposing animal remains in the stratified 
deposits, can only be described as a " Prophet of the Utterly 
Absurd." A fairly full list of palaeoclimatic theories, possible 
and impossible, is given in Appendix II. 

Even when we leave out of consideration the " Patently 
Impossible and Vain," however, there still remain theories 
innumerable. Most of the earlier adventurers in this subject 
had a single cause oceanic circulation, carbon dioxide, 
eccentricity of the orbit, etc. and saw in it complete explana- 
tion of the whole of climatic history. This spirit is not yet 
entirely dead, but of recent years a broader outlook has become 
manifest : 

There are nine and sixty ways 

Of constructing tribal lays, 

And every single one of them is right. 

.There are at least nine and sixty ways of constructing a 
theory of climatic change, and there is probably some truth 
in quite a number of them. The greatest extremes of climate 
are not to be attributed to the abnormal development of any 
one factor, but to the co-operation of a number of different 
factors acting in the same direction. It seems probable that 
the dominant factors have always been " geographical/* but 


we must give this term a very wide meaning, and include not 
only the distribution of land and sea and the systems of ocean 
currents which result from it, but also the vertical circulation 
of the sea, the general elevation of the land, and the amount 
of explosive volcanic activity. It will be seen in ^Chapter XII. 
that all these " geographical " factors are so closely inter- 
related that no discussion which takes account of only one of 
them can be complete. 

But although the results of this work seem to indicate that 
the " geographical " factors have dominated the climatic 
history of the globe, this does not mean that other npn- 
geographical factors are without effect. In Chapter XXII. 
it is shown that during the past two thousand years, when 
the distribution and elevation of the land remained practically 
constant, a number of minor fluctuations of climate have been 
associated with changes of solar activity. No doubt similar 
fluctuations have occurred throughout geological time, but 
they have been masked by the much greater effect of the 
" geographical " factors, and it is only exceptionally that 
solar influences can be recognised, as in the " varve " clays 
formed during the melting of the Quaternary ice-sheets. 
When the " varve " clays of earlier glacial periods, the rings of 
the fossil trees found in various geological horizons, and other 
remains which depend on seasonal changes of temperature 
or rainfall come to be examined in detail, it is to be expected 
that further evidence will be discovered of the existence 
of minor climatic fluctuations which can be attributed 
to small variations of solar radiation. Any such material 
which can be analysed in the way that the tree-rings or the 
Nile flood records have been analysed will be especially 

We know that the more or less regular variations in the 
inclination of the earth's axis, the eccentricity of its orbit, 
and the precession of the equinoxes must have influenced the 
seasons in the past. So far no climatic variations have b^en 
found of such a nature that they can be unhesitatingly 
attributed entirely to these astronomical factors, though it 
is suggested in Chapter V. that they may have been responsible 
for the sequence of deposits in the coal measures and in the 
Tertiary of Southern Europe, and there is good reason to 
believe that the large annual range of temperature in Northern 


Europe during the Boreal or Continental period was partly 
due to a high inclination of the earth's axis. Even in the 
last exVmple, however, the main cause of the extreme climate 
was probably geographical, not astronomical, while the 
historical period in which geographical causes fell into the 
background has not been long enough for any of these 
astronomical factors to change sufficiently to make their 
presence felt. 

Any change, geographical or astronomical, can only become 
effective through its action on the atmosphere, but it will be 
seen that this action is complex, and may at times produce 
unexpected results. During the warm periods, for example, 
there was a large amount of water vapour in the air, yet the 
cloudiness, rainfall, and evaporation were less than at present, 
because the extension of the sub-tropical anticyclones into 
higher latitudes led to more stable conditions. The discussion 
of the effect of cloudiness in Chapter VII. shows that Marsden 
Manson was on the right lines when he emphasised the great 
climatic importance of cloudiness, but he erred in attempting 
to dissociate this element from the pressure distribution and 
the atmospheric circulation in general, and in relating it 
directly to earth heat. His neglect of meteorological principles 
led him into some absurd conclusions, and the same is true 
of many other theorists in the domain of palaeoclimatology. 
The theory which attributed all the great climatic changes to 
variations in the amount of carbon dioxide is a case in point ; 
after a series of highly laborious and intricate discussions from 
the geological side, it was laid gently to rest by the application 
of simple physical experiments, and it is now generally 
recognised that variations in the amount of carbon dioxide, 
while not entirely without some climatic significance, are of 
relatively slight importance. Chamberlin's ingenious theory 
of the reversal of the oceanic circulation to give warm periods 
in high latitudes, although it never reverberated so widely 
thrQugh the world as did the original carbon dioxide theory, 
really has a better physical basis. It is shown in Chapter III. 
that a change in the type of oceanic circulation, although on 
different lines to Chamberlin's, may have been of great 
climatic importance, but only as an auxiliary factor in con- 
junction with favourable geographical and meteorological 
conditions, * 


The second section of the book applies the principles laid 
down in the first section to the various problems presented 
by geological climates. It opens with a compari/on by 
the method of correlation of the climatic history of the 
north temperate and polar regions with the 9 corresponding 
changes of elevation, land and sea distribution, oceanic 
circulation, and volcanic activity. In this way numerical 
measures are obtained of the influence of these various 
geographical factors on geological climates, which are found 
to be in good agreement with the theoretical values deduced 
from recent meteorological work. The next two chapters 
are mainly devoted to a consideration of the theory of 
continental drift. In Chapter XIII. the theory is stated, with 
some reference to its geological basis. Chapter XIV. considers 
the extent to which the theory satisfies the requirements of 
palaeoclimatology, and it is shown that even on Wegener's 
reconstruction of the Upper Carboniferous geography, the 
climate of that period still presents many difficulties. Next, 
in Chapter XV., the distribution of land and sea, mountain 
ranges and ocean currents during the Upper Carboniferous 
is set out on the basis of the present positions of the continents, 
and the probable climate to which such a distribution would 
give rise is deduced from the principles developed in Part I. 
It is found that these give at least as good an approximation 
to the facts as does Wegener's theory, and it is accordingly 
inferred that continental drift is not necessary to account for 
the distribution of past climates. 

I may, I hope, be excused for the length to which the 
discussion of continental drift has run. The theory is at 
present on its trial before the tribunal of the world's scientists, 
and the verdict appears to be wavering in the balance. 
Geology, naturally, will be the final arbiter, but the voices 
of other sciences are not without some weight. Meteorologists, 
thinking always in terms of the well-known distinction that 
while weather changes from day to day climate goes op. all 
the time, are naturally averse to considering the possibility 
of changes in the atmosphere circulation of sufficient magnitude 
to give climatic revolutions on the geological scale. To most 
meteorologists, therefore, Wegener's theory of continental 
drift presents an easy escape from the palaeothermal problem. 
But it seems to me that this point of view is similar to that 


of the captain of a vessel who, crossing the North Atlantic, 
should attempt to determine his latitude with the aid of a 
thermometer and isothermal chart. 

* There seems also to be some confusion of thought as to 
Wegener's theory itself. There is some definite geological 
evidence in favour of that part of the theory which states that 
the alternate stretching and compression of the crust, in 
conjunction with tidal forces, has led to the gradual drifting 
apart of the continents in an east-west direction. This being 
granted, there is a tendency to assume that the remaining 
part of Wegener's theory is also well grounded, namely, that 
which attributes to the continents, in addition to the com- 
paratively small east-west movements, enormous drifts from 
north to south or vice versa. The only real evidence adduced 
in support of this view is climatological, and, practically 
speaking, the climate of the Upper Carboniferous ; the geo- 
logical evidence is quite inadequate. It is therefore necessary 
to examine the climatic evidence carefully from all points 
of view, and especially to explore the possibility of alternative 
explanations to it. Wegener's explanation, though not 
probable, is possible ; in Chapter XV. I have attempted to 
set out an alternative explanation which likewise, though 
not probable, is possible. Which of us is on the right lines, 
time will show. But even if the theory of continental drift 
should ultimately be established in all its parts, the necessity 
for a science of " palaeometeorology " will still remain. It 
is now quite beyond doubt that the earth has passed through 
periods of mountain formation alternating with long periods 
of comparative rest, and that in consequence the average 
elevation of the land has varied considerably from time to 
time. Similarly, on any theory (and on Wegener's theory 
more than on any other), the distribution of land and sea has 
varied greatly from one period to another. These changes 
must inevitably have produced corresponding changes in the 
distribution of climate, which can only be arrived at by 
applying the principles of meteorology . Hence it is hoped that 
this book will be of service to geologists and palaeoclimatologists 
no matter what basal theory of past geography they may 

The third section of the book deals in considerable detail 
with the climates of different pal ts of the world during the 


historical period, or from about 5000 B.C. to the present day. 
In recent years a large amount of valuable work has been 
published dealing with the climatic changes during this 
period, and this part of the book is materially more complete* 
and definite than would have been possible had it been written 
even as little as ten years ago. I have been fortunate in being 
able to make use of a number of detailed and entirely in- 
dependent records for different parts of the world, such as the 
annual rings of the big trees of America, the literary and 
historical records of Europe and of China, the levels of the 
Caspian, the racial movements of Asia, and the floods and 
low-water stages of the Nile, and these have shown so good 
an agreement with each other and with such records of solar 
activity as we possess, that I cannot but feel that the climatic 
fluctuations portrayed are definitely real and demonstrate the 
solar control of climate in the absence of disturbing 
geographical changes. 

Apart from its meteorological interest, I hope that this part 
of the work will prove of service to archaeologists and 
historians. Mr Harold Peake, in his brilliant study of " The 
Bronze Age and the Celtic World," found it necessary to call 
in the aid of climatic changes in order to understand the 
migrations of the " Aryans," which laid the foundations of 
the present distribution of peoples in Europe and Western 
Asia. Similarly, Ellsworth Huntington believes that the 
rise and decline of the ancient Maya civilisation of Yucatan 
can only be explained by changes in the climate, and there 
are other examples. The literature of historic climates is, 
however, so chaotic that the historian or archaeologists has little 
inducement to trust himself to its mazes. A co-ordination 
of the evidence was urgently needed, and this need I have 
attempted to meet. 

I wish to make acknowledgment to the Council of the 
Royal Meteorological Society for permission to reprint parts 
of several papers published in the Quarterly Journal, including 
Figures i, 8, and 16 to 19, for several of which they have 
kindly lent the blocks. Mr W. H. Dines, F.R.S., has kindly 
accorded me permission to use his illustration of the heat 
balance of the atmosphere (Fig. n). Messrs G. W. Bacon & 
Co. have accorded permission for the use of their outline chart 
of the globe on Mercator'^ projection as a basis for several of 


the charts. Several friends have read parts of the manuscript 
and have made valuable criticisms and suggestions, or have 
assisted me with expert advice on different aspects of a subject 
which is so wide that no one man can hope to master it. 

C. E. P. B. 

September 1926. 


The twenty-two years which have elapsed since the pre- 
pafation of the first edition have added to the literature, 
among other works, Sir George Simpson's theory of variations 
of solar radiation and Professor Zeuner's brilliant exposition 
of the astronomical causes of the succession of glacial and 
interglacial periods ; the explanation of the glacial succession 
in the Quaternary now seems to rest with one or possibly both 
of these theories. Neither of them accounts for the occurrence 
of the Ice-Age as a whole or for the warm periods, and with 
the piling up of objections to Wegener's hypothesis of con- 
tinental drift, the case for the " geographical " theory has 
been strengthened. There has been a general acceptance 
of the idea of " glacial " and " non-glacial " climates which 
forms the basis of that theory. In the post-glacial period the 
principal changes of the past twenty years have been a new 
conception of the climate of the Sub-boreal and the dating 
of the beginning of the Sub-atlantic as 500 B.C. instead of 
850 B.C., both of which make the post-glacial sequence more 
intelligible. It is now possible to present a much more 
complete interpretation of the changes of climate in the 
historical period than could be given in 1926. 

While the general plan of the book remains unchanged, 
several chapters have been almost entirely rewritten. It 
seemed unnecessary to reprint the Appendix describing the 
mathematical theory of correlation, which is now widely 
known and is available in many text-books on Statistics. 
My thanks are due to the many authors who have sent me 
copies of their publications dealing with climatic changes, 
which have saved me much arduous search in libraries, to 
Sir George Simpson and the Manchester Literary and 
Philosophical Society for permission to reproduce Figures 
9 and 10, to the publishers for bringing out a new edition in 
the face of great difficulties, to Miss N. Garruthers for reading 
the revision and making a number of valuable suggestions, 
and especially to my wife for her great practical help and 
encouragement during the course of the revision. 

C. E. P. B. 
February 1948. 



















INTRODUCTION. The Normal Climate of Geological Time . 2 1 


" Glacial " and " Non-Glacial " Periods . 31 
Pressure and Winds .... 46 
The Circulation of the Oceans . . 68 
Radiation from the Sun .... 89 
Astronomical Factors of Climate . . 102 

VI. The Absorption of Radiation by the Atmo- 
sphere . . . . . .no 

VII. The Effect of Cloudiness on Temperature 122 

Chapter VIII. Continentality and Temperature . . 129 

Chapter IX. Precipitation Rain, Snow, and Hail . 1 58 

Chapter X. Mountain-building and Climate . . 177 

Chapter XI. The Weather of the Warm Periods . . 192 

Chapter XII. The Geography of the Past . . .201 
Chapter XIII. The Theory of Continental Drift _ . . 221 

Chapter XIV. An Examination of the Climatic Evidence 

for Continental Drift . . . .231 

Chapter XV. The Climate of the Upper Carboniferous 

Glacial Period 247 

Chapter XVI. The Climate of the Quaternary . . 263 

Chapter XVII. The Nature of the Evidence . . .281 
Chapter XVIII. Europe . .... 295 



Chapter XIX. Asia 318 

Chapter XX. Africa 329 

Chapter XXI. America and Greenland .... 342 

Chapter XXII. The Interpretation of Climatic Fluctuations 

in the Historical Period . . . 359 

APPENDIX I. The Geological Time-Scale . . . 379 

APPENDIX II. Theories of Climatic Change . . . 384 

INDEX ........ . 387 



Fig. i. Cooling at edge of floating ice-cap . . . 37 

Fig. 2. Temperature difference between " non-glacial " 

and " glacial " climate ..... 44 

Fig. 3. Mean pressure, January ..... 47 

Fig. 4. Mean pressure, July ...... 48 

Fig- 5- Winds over North Pacific, January . . . 57 

Fig. 6. Winds over North Pacific, July .... 58 

Fig. 7. Reconstruction of pressure and winds during a 

warm period ....... 62 

Fig. 8. Ocean currents in winter ..... 70 

Fig- 9- Effect of increasing radiation on precipitation 

and accumulation of snow . . . . 92 

Fig. 10. Effect of two cycles of solar radiation on glaciation 93 

Fig. 1 1 . Heat exchange of atmosphere . . . 1 1 1 

Fig. 12. Average temperature distribution, January . . 130 

Fig. 13. Average temperature distribution, July . . 131 

Fig. 14. Isotherm of 32 F. in Tertiary and Quaternary . 135 

Fig. 15. Change of temperature due to formation of an 

island . . . . . . . .145 

Fig. 1 6. Observed, and calculated temperature changes, 

Litorina period, January . . . . .147 

Fig. 17. Changes of level and land and sea distribution, 

Quaternary . . . . . . .152 

Fig. 1 8. Changes of temperature due to geographical 

changes, January . . . . . 154 

Fig. 19. Changes of temperature due to geographical 

changes, July 155 

Fig. 20. Tracks of depressions 160 

Fig. 21. Variation of rainfall with latitude . . .165 

Fig. 22. Relations of soil to temperature and rainfall 167 




Fig. 23. Mountain-building and glaciation (schematic) . 178 

Fig. 24. Geographical factors and temperature during 

geological time ...... a 06.' 

Fig. 25. The transition from a warm period to an 'ice-age . 218 

Fig. 26. Carboniferous and Permian deserts . . . 237 
Fig. 27. Mesozoic deserts . . . . . . .238 

Fig. 28. Land and sea distribution, Cretaceous . . 240 

Fig. 29. Geography of the Upper Carboniferous . . 248 

Fig. 30. Ocean currents of Middle Permian . . . 252 

Fig. 31. Variations of rainfall in Europe .... 299 

Fig. 32. Variations of temperature in Europe . . . 311 
Fig* 33- Variations of rainfall in Asia . . . .321 

Fig. 34. Variations in the level of the Caspian . . . 323 

Fig- 35- Levels of Nile. Flood stage and low-level stage . 330 

Fig. 36. Variations of rainfall in Africa .... 339 

Fig. 37. Variations of rainfall in U.S.A. .... 343 

Fig. 38. Variations of rainfall, world .... 359 

Fig. 39. Causes of climatic variations since 6000 B.C. . 361 



DURING some hundreds of millions of years with 
which we are acquainted through the records of 
the rocks, the surface of the earth has passed through 
some strange climatic vicissitudes. At least four, probably 
five or more times in its history great ice-sheets have spread 
out from various centres, covering the plains and even filling 
the shallow continental seas. There have been other periods 
of somewhat less intense climatic stress, when perhaps only 
small glaciers were able to develop among the high hills. 
These strenuous episodes have been of great importance in the 
development of the earth's living beings, but always they have 
been brief, and always after them the earth has returned to 
more genial conditions, which have endured for long periods. 
Hence we may regard these genial conditions as the normal 
state of affairs on this our earth, and the glacial periods as 
episodes disturbing the normal climate for a brief time, as at 
long intervals a passing cyclone disturbs the peaceful life of a 
tropical island. Many references to these warm periods will 
occur in the next few chapters, and it is necessary to have 
some preliminary knowledge of the climatic conditions which 
characterised them. I am therefore beginning this study, 
not with the stir and strife of an ice-age, but with the everyday 
life of the genial periods. We are not yet in a position to 
discuss their meteorology, but we can see what the geologists 
have to say about their general climate. 

The first point to notice about these periods, however, is not 
climatic, although we shall see later that it has a very great 
beaming on their climatology ; it concerns the absence of 
mountain ranges. Warm periods were without exception, 
periods of low relief, while glacial periods occurred when the 
earth's crust had been thrown into folds and ridges by great 
internal convulsions. The processes of denudation wave 
action on the coasts, running water on the slopes and in the 
valleys, and blown sand in the %teppes and deserts go on 


all the time, and are constantly tending to level up the earth's 
crust, and they are powerfully aided during the cold periods 
by the action of frost and ice on the high ground. Against 
them are set the internal mountain-building forces, whfch 
act with great intensity for comparatively sh'ort periods at 
great intervals. The forces of denudation are most powerful 
when the land is highest ; following a period of mountain- 
building the general level is reduced very rapidly at first, and 
the normal stage of low relief is soon reached again. 

During the periods of low relief also the sea generally 
encroached on the land and the continental areas were greatly 
reduced in size. This we shall see was another factor of 
great climatic importance, first, because it did away with the 
areas of intense winter cooling in the centres of the great 
continents in middle to high latitudes ; and secondly, because 
the extension of the seas allowed the warm ocean currents to 
penetrate very readily into the polar regions through many 
broad channels. This gives us the background for our 
climatological study groups of broad low islands rather than 
continents, with rounded hills instead of mountains, and wide 
oceans extending through broad channels from pole to pole. 

The next point is the comparatively small difference of 
temperature between the tropical and the polar regions. 
The marine fauna, and to a less extent even the land vegetation, 
differ little whether we are in Greenland, in Yorkshire, or in 
Central America. When the rich fossil faunas and floras 
from high latitudes were discovered in the middle of the 
nineteenth century and this similarity in the life from widely 
different latitudes was first remarked, it was believed that 
during the early geological periods there were in fact no 
biological zones at all, and that life was really uniform over the 
whole extent of the seas. Radio-activity had not been heard 
of, and it was believed that the high temperatures of the earth's 
interior indicated by volcanic eruptions were mainly a legacy 
from the time, not long before the Cambrian, when the (jarth 
was entirely molten. It was believed that for a long ime 
after the formation of a solid crust, the condensation of water 
as oceans, and the origin of life, this internal reservoir of heat 
continued to make itself felt at the surface, and maintained 
genial temperatures in all latitudes throughout the Palaeozoic, 
the Mesozoic, and a large part of the Tertiary. The 


pre-Cambrian and Carboniferous glaciations were unknown, and 
the apparently uniform, equable temperatures of these earlier 
periods fitted excellently into this theory of a cooling earth. 
It is now recognised, however, that the earth has owed its 
surface warmth to the sun as far back as we can see into the 
past, and that even in the warmest periods there must have been 
zonal differentiation of climates ; the tropics were perhaps 
a little warmer than they are to-day, while the polar oceans 
and their shores had a temperature now found in temperate 
latitudes. Since the deposits which have come down to us 
were almost invariably formed in the sea or in the estuaries 
of wide rivers, we know comparatively little of the plants which 
grew in the interior of the continents in high latitudes ; what 
we do know suggests that they were of hardier types than those 
from the low coastal valleys. 

Neumayr ( i ) was the first to bring forward definite evidence 
of a zonal distribution of animals in earlier geological periods ; 
his conclusions, as revised by V. Uhlig (2), distinguish five 
faunal zones during the Jurassic : Boreal, Mediterranean- 
Caucasian, Himalayan, Japanese, and South Andean but 
the four latter all seem to be facies of a tropical zone which 
is contrasted with the cooler boreal. The boundary between 
these zones does not run strictly parallel with the lines of 
latitude, but shows divagations which may reasonably be 
attributed to ocean currents. Similarly, during other 
geological periods, the faunas of different latitudes when 
examined critically invariably show differences between 
different latitudes which are best accounted for by a slow 
decrease of temperature towards the poles ; for example, 
during the Cambrian, the Archaocyathim of the Antarctic 
show the same species as those of Australia, but in a dwarfed 
and crippled condition. In the Lower Cretaceous we have 
a clear division of the marine faunas into northern boreal, 
Mediterranean-equatorial, and southern boreal, the latter 
containing some species identical with those in the first, but 
absent in the second. The principal difference between the 
temperate and the polar faunas throughout the Mesozoic is 
the absence or dwarfing of the corals in the latter. 

Similarly, although a rich vegetation apparently extended 
as near the present poles as the land surfaces permitted, 
there are considerable differences* between the lower Tertiary 


floras of the sub- Arc tic regions and those of lower latitudes. 
Thus E. W. Berry (3) records " the total dissimilarity between 
the Canadian floras, which are a part of the [Upper Eocene] 
Arctic flora of Alaska, Greenland, Iceland, Spitsbergen, etc., 
and the contemporaneous flora of the [Mexican] Gulf States." 
The Arctic flora is found also in Northern and Eastern Asia, 
so that its distribution covers an irregular area surrounding 
the pole, the southern limit varying from 45 N. to the Arctic 
Circle. The plants include poplar, willow, alder, birch, 
hazel, beech, oak, plane, laurel, andromeda, ash, guelder 
rose, cornel, magnolia, ivy, spindle-tree, buckthorn, sumach, 
and hawthorn. All these are now found in some parts of the 
cool temperate zone, and the assemblage, while presenting a 
very different picture from the present life of the sub-Arctic 
regions, cannot be described as sub-tropical. Berry considers 
that the identifications of palms, etc., which have given rise 
to the ideas as to very high polar temperatures in the early 
Tertiary, are erroneous. The most northerly Eocene flora 
at present known is from near Cape Murchison in Grinnell 
Land, latitude 71 55' N. ; Berry reduces the list of Heer's 
determinations of this flora to the following : horse-tail, 
yew, pine, spruce, poplar, birch, hazel, sedge or grass, and 
apparently water-lily but he remarks that even this flora is 
sufficiently remarkable when we consider the scantiness of the 
present vegetation of the region, and would require the 
present isotherms to swing 15 or 20 northward. 

Not only has the northern boundary of this cool temperate 
flora moved southward since the Eocene, but so also has its 
southern boundary, the shift being about 10 of latitude. 
This suggests that the tropical as well as the polar regions 
were warmer in the Eocene than they are at present, though 
not to the same extent, and that definite climatic zones were 
in existence at that time. 

The third point to notice about the warm periods is their 
general aridity. Deserts have apparently existed throughout 
geological time, but during most of the warm periods, and 
especially in the Mesozoic, they expanded greatly, extending 
from the sub-tropical regions far into the present temperate 
zones. Among the most remarkable of these " fossil deserts " 
we have the " Old Red Sandstone " of the Devonian period, 
consisting mainly of inlancf lake or lagoon deposits formed 


under arid or semi-arid conditions. This regime extended 
from Ireland across Britain to Southern Norway, Poland, 
Cgurland, and the White Sea, and similar deposits are found 
in Nova Scoti^, New Brunswick, Canada, and the north-east 
of the United States. This region was not altogether rainless, 
but the rain came in occasional heavy showers which caused 
brief floods in the water-courses, carrying large quantities of 
coarse detritus into the lakes and depressions. The desert 
character of the land was accentuated because the specialisation 
of the plants for life on dry land had scarcely begun, and it is 
quite likely that regions with a similar climate at the present 
day would have a moderately rich vegetation. They have 
therefore been described as biological rather than climatic 

Desert deposits are again well developed in the Permian, 
in Britain, France, Germany, and the Tyrolese Alps. During 
this period a great inland sea was formed over a large part of 
Europe ; since evaporation exceeded rainfall this sea became 
highly saline and deposited thick layers of salts. These show 
in part regular annual layers, gypsum, which is more soluble 
in cold than in warm water, being formed in summer, while 
in winter its place was taken by rock salt. Kubierschky, 
quoted by Koppen and Wegener, on the basis of similar 
phenomena in the Sahara, estimates the annual range of 
temperature as between 60 and 95 F. The number of annual 
layers indicates that the salt lake existed for some 10,000 years, 
after which the salt deposits were covered by a layer of desert 

In the Triassic we have a very wide development of desert 
formations, especially in Western and Central Europe and in 
the east of the United States, Texas, Colorado, and Idaho. 
Many of these may have been biological rather than climatic 
deserts, but the occurrence of numerous salt deposits, especially 
in Central Europe, shows that the rainfall was less than the 
evaporation over wide areas. In the Jurassic and Cretaceous 
there is less evidence of extensive deserts, partly because the 
climate was not so arid as that of the Triassic, but partly 
because the land vegetation had evolved more specialised 
types which were able to exist without the constant presence 
of water. Occasional salt deposits show, however, that the 
climate of Europe was still rather* dry. During the Tertiary, 


the Eocene period shows a return of moister conditions, but 
the Oligocene and, to a less extent, the Miocene, were again 
dry over much of Europe. The Mediterranean regions had 
for a time a true arid climate, while Egypt wjis for long an 
absolute desert, but in Central Europe, although the summer 
was hot and dry and was prolonged into autumn, there was 
heavy rainfall of the thunderstorm type in spring, which tore 
leaves and twigs from the trees and bushes and carried them 
to the lakes, where they were buried in mud. The winter 
was generally mild, and there were many evergreens, but some 
of the Miocene leaves from Central Europe show traces of 
frost action, indicating that there were occasional cold nights. 
Generally speaking, the climatic zones in Europe lay 10 or 
15 degrees north of their present position. 

In Western U.S.A., where the succession of events has been 
worked out in great detail, the Eocene and early Oligocene 
enjoyed a sub-tropical climate with ample rainfall, becoming 
temperate farther north in Alaska. In middle Tertiary the 
climate became increasingly cooler and more arid, especially 
to the east of the Rocky Mountains, this process culminating 
in the cool semi-arid climate of the Pliocene. 

It is probable that one of the most momentous steps in the 
history of life was taken during a period of drought, namely, 
the origin of air-breathing land vertebrates. A similar 
development on a smaller scale was the evolution of the Lung- 
fish (Ceratodus], adapted to breathe both air and water, which 
to-day inhabits regions subject to alternating floods and 
droughts. Ancestral forms appeared in the Devonian, and 
the true lung-fish as early as the Permian ; it was widely 
distributed in the Triassic of Central Europe. 

Thus it will be seen that the predominant features of the 
normal geological climate were warmth and dryness. Broadly 
speaking, the polar regions had the climate of the present 
temperate belts, while the latter had the climate of the sub- 
tropics. In the latitude of the British Isles the rainfall oame 
almost entirely in the form of brief, heavy showers, of the type 
associated with thunderstorms, and the steady but gentle rains 
which are characteristic of our winter months did not occur. 
These rains are associated with the passage of extensive 
barometric depressions or cyclones, and from their absence 
we can infer that during ttc warm periods such depressions 


either did not occur, or were limited to high latitudes. Thus 
the meteorology of these warm periods was very different 
from that of to-day, and it is these differences which we have 
to analyse and, if possible, account for, in the following pages. 


(1) NEUMAYR, M. " Uber klimatische Zonen wahrend der Jura- und Kreidc- 

zeit." Wieri, Denkschr. K. Akad. Wiss., Math. nat. KL, 47, 1883, p. 277. 

(2) UIILIG, V. " Die marinen Reiche des Jura und der Unterkreide." Wien, 

Mitt. Geol. Ges., 4, 1911, p. 329. 

(3) BERRY, EDWARD W. " A possible explanation of Upper Eocene climates. " 

Proc. Amer. Phil. Soc. 9 61, 1922, p. i. 




IN the preceding Introduction we saw that one of the 
characteristics of the " warm " periods which have formed 
the major part of geological time was the small temperature 
difference between equatorial and polar regions, small, that 
is, relative to the present difference. This at once introduces 
a difficulty which the meteorologist feels very acutely in 
attempting to account for geological climates, for the basis 
on which the existence of the major climatic zones depends 
is not to be found in any surface features of the earth, which 
might have been different in some former geological period, 
but in the fact that the earth is very nearly spherical. So 
long as the axis of rotation remains in nearly its present position 
relative to the plane of the earth's orbit round the sun, the 
outer limit of the atmosphere in tropical regions must receive 
more of the sun's heat than the middle latitudes, and the 
middle latitudes more than the polar regions ; this is an 
inviolable law. It is not difficult to think of causes, such as 
a change in the heat of the sun, or the formation of a veil of 
volcanic dust in the atmosphere, which will slightly raise or 
lower the mean temperature of the earth as a whole, or change 
the temperature of equatorial regions more than that of the 
polar regions ; it is much more difficult to think of a cause 
which will raise the temperature of polar regions by some 
30 F. or more, while leaving that of equatorial regions almost 
unchanged, and so bring about an approach to the distribution 
of climatic zones during the warm periods. 

Let us consider what is the essential difference between 
polar and equatorial regions to-day. We can say that the 
mean temperature of the polar regions is 20 F., and that 
of the equatorial regions 80 F., and the difference between 
20 and 80 is very striking. But the zero on the Fahrenheit 
scale is purely arbitrary ; the true zero of temperature is at 
459 F., or to adopt the units employed by physicists, 273 
C., and if we express these temperatures on the absolute 


scale the figures become polar, 266 A, equatorial, 299 A, 
and the difference does not look so alarming. The essential 
difference is not between 20 and 80 F., but between the fact 
that the former is below the freezing point of water and ihe 
latter is above it. The real difference betweeh the centre of 
Africa and the centre of Antarctica is that in the former, 
water is water and in the latter, water is ice. Similarly, the 
essential way in which the polar regions during the warm 
periods differed from the polar regions at present was that 
they had no great ice-sheets and no floating ice on the sea, that 
water was water, and they were " non-glacial." 

The winter cold of Siberia is well known, and is explained 
in the geographical text-books by the statement that the 
climate is " continental." With the coming of the short days, 
in which the sun has little heating power, and the long nights, 
the surface of the ground loses its heat very rapidly, and soon 
falls below freezing point. At a depth of a foot or two the 
soil may be much warmer, but heat passes very slowly through 
the ground, and this underground store of heat does not help 
appreciably to keep up the temperature of the surface. Soon 
snow falls on the frozen ground ; snow, too, is a very poor 
conductor of heat, and moreover its white surface reflects 
four-fifths of the sun's heat which falls on it ; it absorbs heat 
very slowly and loses heat very readily, hence the temperature 
falls still more rapidly, until life becomes barely supportable. 
Compare this with the conditions in the centre of the Pacific 
Ocean in the same latitude. As the surface layer of water 
cools, the winds and waves, aided by convection, mix it with 
the underlying water, and in this way the cooling which was 
practically limited to a few inches of soil or snow over the 
land, becomes spread through many feet of water in the oceans. 
This means that the sea surface cools very much more slowly 
than the land, quite apart from the fact that water has a higher 
heat capacity than soil or rock. But when water freezes and 
loses its mobility, it takes on some of the properties of land, 
and when snow falls on the ice, we have a surface which is 
indistinguishable from the surface of the snow-covered land. 
Thus an Arctic Ocean covered by a snow-encrusted layer of 
floating ice acts almost as a northward extension of the con- 
tinents of Asia and America. The temperature does not fall 
quite so low as in Siberia, partly because, the surface being 


more level, the cold air can escape more readily, and partly 
because a little more heat finds its way from below through 
ice than through the subsoil, but a temperature of 62 F. was 
recorded over the ice in March 1894 during Nansen's " Fram " 

But the cooling power of a snow-covered surface is not 
limited to the surface itself; the winds spread its influence 
over the surrounding land or ocean. The winds blowing 
off the snow of Canada lower the mean temperature of the 
whole of North America except the Pacific coast, and this 
cooling allows the snow-cover to extend much farther south 
than it would do if, for instance, these cold winds were held 
off by a range of mountains extending from east to west. 
Investigations to be described in detail later (Chapter VIII.) 
have shown that if in the middle of a large open polar ocean 
a circular cap of ice were formed, with its centre at the pole 
and its southern edge in latitude 80 N. (that is, having a 
radius often degrees of arc or about 690 miles), the temperature 
would be lowered by about 45 F. at the pole and about 
20 F. in latitude 80 on the edge of the ice, while the cooling 
would still be quite appreciable at a distance of 500 miles 
south of the ice-edge, across the open ocean. 

Now let us suppose (after Brooks (i)) that the earth is 
entirely covered by water for a distance of more than 2,000 
miles from the North Pole, and further suppose that the surface 
of the water at the pole itself is just warm enough to prevent 
freezing, while the temperature rises outwards from the centre 
at a uniform rate. Now suppose that a small uniform decrease 
of temperature occurs over the whole ocean. This results in 
the formation of a mass of floating ice in the centre, which 
will extend to the limit of the area reduced below the freezing 
point by the initial fall of temperature. This ice itself exerts 
an additional cooling effect on the air, not only above the ice, 
but for some distance round it, so that a further area of the 
sea surface has its temperature reduced below freezing point 
and becomes converted into ice. The ice-cap accordingly 
extends beyond the limits of the area of freezing due directly 
to the initial fall of temperature. Equilibrium is reached only 
when the initial decrease of temperature, plus the cooling effect 
exerted by the ice-cap at a point on its edge, is balanced by the 
rise of temperature due to increasing distance from the pole. 


Now we come to the consideration that while the rise of 
temperature due to the horizontal temperature gradient is 
proportional to the distance from the pole, the area of the 
ice-cap increases with the square of its radius. Consequently, 
if the cooling effect of an ice-cap at a point on its edge varies 
directly with its area, the cooling effect is small at first, but 
increases more and more rapidly as the ice-cap grows in size, 
while the counteracting effect of the normal horizontal 
gradient increases at a uniform rate. Sooner or later a point 
is reached at which a further growth of the ice-sheet causes a 
greater lowering of temperature on its edge than can be 
neutralised by the horizontal temperature gradient, and beyond 
that point, in the conditions postulated, the ice-cap must 
continue to grow indefinitely. The critical point at which 
the ice-cap becomes unstable depends on the numerical values 
assigned to the cooling power of ice and to the horizontal 

This conclusion is so important, underlying as it does the 
whole theory of climatic changes as set out below, that I may 
be excused for dwelling on it at some length. First, is the 
cooling power of an ice-cap on its edge proportional to its 
area ? The winter cooling by floating ice is quite comparable 
to the winter cooling due to the presence of land in high 
latitudes, since in winter the land is almost invariably snow 
covered. It is shown in Chapter VIII. that with increasing 
land area, the cooling effect exerted by each square kilometre 
of land in winter actually increases at first with the increase 
in total area, so that by doubling the area we more than 
double the cooling. In the centre of a round island the 
maximum cooling effect per square mile of total area occurs 
when the island has a radius of about ten degrees of arc ; 
beyond this the more distant parts cease to exert their full 
effect, and while the total cooling effect continues to increase 
with increasing area, the average effect per square mile begins 
to decrease. On the edge of a large island the maximum 
effect is exerted only by that portion which lies within a circle 
of ten degrees radius centred at a point on the edge ; when the 
radius of the island exceeds eight degrees, any further increase 
of radius makes a comparatively small difference in the area 
within this ten-degree circle. We shall not be far out if we 
assume that the fall of teniperature due to the ice at a point 


on its edge is directly proportional to the area of ice included 
in a circle of ten degrees radius, drawn round that point, 
everything outside that circle being ignored. As the average 
effect per square mile of land between ten and twenty degrees 
away is only one-fifth of the average effect of land inside the 
circle of ten degree radius, this simplification is justified. 

The next point is to determine the critical temperature 
given by definite values of horizontal gradient and cooling 
power of ice. We still start with a simple numerical example. 
Let us put the cooling power of a floating ice-cap as 0-5 F. 
for each one per cent, of a circle of ten degrees radius which 
is occupied by ice. Then on the edge of the ice-cap the 
lowering of temperature due to the ice will be 0-5 F. when it 
has a radius of one degree ; four times 0-5 or 2-0 F. when 
it has a radius of two degrees; nine times 0*5 or 4-5 F. 
when it has a radius of three degrees, and so on. We will 
suppose that before the ice-cap developed the temperature 
increased uniformly outwards from the pole at the rate of i F. 
for each degree of latitude. Then, taking the freezing point of 
sea water as 28 F. and supposing that initially the sea at the 
pole was just on the point of freezing, we have the following 
distribution of temperatures before the formation of the ice- 
cap and at different stages of its growth : 

Distance from pole, degrees 

of latitude o i 23 45 

Initial temperature, F. . 28 29 30 31 32 33 

Cooling due to ice-cap, F. o 0*5 a 4-5 8 12-5 
Temperature on edge of 

ice-cap, F. . . . 28 28*5 28 26-5 24 20-5 

Table i . Temperature at edge of floating polar ice-cap. 

This table shows that the floating ice-cap will experience most 
difficulty in establishing itself; once it has reached a certain 
size the temperature on its edge will be below the freezing 
point of sea water, and it will continue to expand owing to 
the lowering of temperature which the ice itself introduces. 
Suppose that, starting with a temperature of 28 F. at the pole, 
we have a general fall of temperature by 0-5 F. An ice-cap 
will form and will grow until it has a radius of one degree of 
latitude. At this stage the temperature at its edge, at a distance 


of one degree from the pole, which was originally 29 F., will 
have been lowered 0-5 F. by the initial cooling and a further 
0-5 F. by the ice, a total decrease of i F. The water on 
the edge of the ice will therefore be exactly at freezing point ; 
lower the general temperature a little more, and the ice-cap 
will continue to grow until it reaches very large dimensions ; 
lower it a little less, and the ice will extend less than one 
degree from the pole. This result is not dependent on the 
special numerical values which have been taken. Suppose 
for instance, that the cooling power of ice is only 0-25 F. 
for each one per cent, of a ten-degree circle, instead of 0-5 F. 
as before. Then the critical radius of the ice-cap is four 
degrees of latitude instead of one degree, and for it to grow 
to large dimensions the initial lowering of temperature must 
exceed i F. instead of 0-5 F. 

Let us now put the matter quite generally. Let the horizontal 
gradient of temperature be h degrees per degree of latitude 
and let the cooling power of ice be k degrees for each one per 
cent, of a circle with a radius of ten degrees. An initial small 
fall of temperature by / degrees will cause the formation of an 
ice-cap with a final radius greater than t/h ; call this radius 
R, where R is measured in degrees of latitude. The cooling 
power of this ice-cap is A:R 2 , and in a condition of equilibrium 
the total cooling t+kR 2 is balanced by the horizontal increase 
of temperature AR. Thus we have the equation : 

the solution of which is 

When =o, R=o, so that only the negative value of the root 
term is required here. The critical point occurs when 
4^= h 2 , or th 2 l^k ; for values of t above this amount the 
equation has no solution, and on the assumptions made the 
ice-cap has no finite limit. 

The average value of the horizontal gradient of air 
temperature in January (K) which would prevail over an ice- 
free ocean between 80 and 90 N. latitude is difficult to 
estimate, but would probably be of the order of 0-5 C. 
(0-9 F.) per degree of latitude. The cooling power of land 
in winter in high latitudes is approximately 0-5 C. (0-9 F.) 



for one per cent, of a circle with a radius of ten degrees ; the 
cooling power of a continuous surface of thick ice would be 
nearly the same, but where the ice is interrupted by lanes of 
open water caused by ocean currents the cooling power is less, 
and we may take k as 0-25 C. (0-45 F.) for one per cent. 
Hence the critical value of t for which the ice-cap becomes 
unstable is 

h 2 0-81 

4# 4 x o 45 

at which stage the radius R is one degree of latitude. We 
can now calculate the total cooling on the assumption that 
the initial fall of temperature is o 6 F. and that ice distant 
from any point more than ten degrees of arc has no influence 
on the temperature of that point. 
The results are as follows : 

Radius of ice- 
cap R, in . o i 2 3 4 5 6 8 10 15 20 25 

Total cooling 
on edge, F. 0-6 1-05 2-4 4-6 7-8 n-8 14-5 17-2 18-2 20-4 21-3 22-2 

Rise due to 
normal hori- 
zontal grad- 
ient A, F. . o-o 0-9 1-8 2-7 3-6 4-5 5-4 7-2 9-0 13-5 i8'O 22-5 

Table 2. Cooling at edge of ice-cap. 


^ 5 

^ 10 







Fig. i . 

10 15 20 25 

of Ice Cap in Degrees 

-Cooling at edge of floating ice-cap. 


These results are shown graphically in Fig. i, in which the 
curve represents the cooling on the edge of the ice, while the 
inclined straight line represents the normal horizontal gradient. 
Over almost the whole of th^ diagram the curved line 


representing cooling is below the straight line representing the 
normal rise of temperature with increasing distance from the 
pole, and the two do not meet until the ice-cap attains a radius 
of nearly twenty-five degrees of latitude. Hence an inifial 
winter cooling to 0-6 F. below the freezing point will result 
in the formation of a floating ice-cap with a radius of nearly 
twenty-five degrees. 

We have now to consider to what extent this ice-cap will 
melt in summer. Let us suppose that the floating ice-cap has 
been formed in winter owing to a depression of the temperature 
to 24 F. or 4 F. below the freezing point of sea water. When 
summer comes, there is a general warming up ; let us suppose 
(in accordance with a calculation by F. Kerner (2)) that under 
normal or " non-glacial " conditions the summer temperature 
would be 13-5 F. above that of winter, or 9-5 F. above the 
freezing point. The edge of the ice-cap, which was just at 
freezing point in winter, will now rise above freezing point, 
and the ice will begin to melt. But although the " non- 
glacial " temperature, even at the pole, is now above the 
freezing point of water, this does not mean that the ice-cap will 
necessarily entirely vanish. As the ice melts back towards 
the pole, its cooling effect at a point on its edge decreases 
slowly very slowly at first and on the edge of an ice-cap of 
ten degrees radius it is only 4 F. less than on the edge of an 
ice-cap of 25 degrees radius. But the natural or " non- 
glacial " temperature decreases towards the pole at the rate 
of O'9 F. for each degree of latitude, and in a distance of 
15 degrees or 1,035 m il es the temperature will have fallen 
from this cause by 13-5 F. This is exactly equal to the 
9-5 F. by which the summer temperature rises above the 
freezing point, plus the 4 F. due to the decrease in size of the 
ice-cap, and by the time the ice-cap has melted back to a 
distance of ten degrees from the pole, the temperature at its 
edge will again be at freezing point and it will be unable to 
melt back any farther. Hence, so long as the amount <?f the 
summer warming does not exceed the greatest difference 
between the curved line and the straight line in Fig. i, the 
ice-cap will not entirely disappear even in summer ; instead 
a core of ice will remain from which the ice-cap will spread 
out again in the following winter. It is evident that this 
central core will in time cdme to consist of very thick old ice, 



while the outer region in which melting takes place in summer 
will consist only of thinner ice of one season's growth. 

In order that the ice-cap may entirely disappear in summer, 
{Tie summer temperature must rise so far above the freezing 
point that the curved line in Fig. i, representing the cooling 
at the ice-edge, lies entirely above the straight line representing 
the normal horizontal gradient of temperature. From Table 
i, we see that these lines are 10 F. apart when the ice-cap 
has a radius of eight degrees, and in order that the ice may 
disappear in summer, the summer temperature must rise above 
the freezing point by at least 10 F. Allowing an annual 
range of i3*5F., this means that the winter temperature 
must be above 24-5 F. 

Fortunately this hypothesis as to the conditions under which 
an extensive floating ice-cap develops does not rest entirely 
on theoretical arguments. The assumptions made represent 
very fairly a generalisation of the Arctic Ocean, and we can 
confirm the conclusions reached by reference to the existing 
ice-conditions in that ocean. In winter, with the exception 
of an embayment west of Norway kept open by the Gulf 
Stream Drift, the whole ocean north of the Arctic Circle is 
covered by floating ice, much of which breaks up in summer. 
Inside the outer or winter limit is another or summer limit 
representing the area over which there exists a solid mass of 
very thick ice (the " Palaeocrystic Ice ") which never melts, 
winter or summer. The limit of this inner ice-cap has an 
average latitude of about 78 degrees, or 12 degrees from the 
pole, and this may be taken as representing the limit of the 
region of maximum depression of temperature below freezing 

It may be remarked in parenthesis that this theory of a 
critical temperature applies equally well to the case of an 
ice-sheet over land, and affords a satisfactory explanation 
of the rapid growth and decay of the Quaternary ice-sheets 
at certain stages of their existence. 

The critical point for an ice-sheet on a conical continent 
with a surface slope of i in 1,000 (e.g., height 2,000 metres, 
radius 20 degrees of latitude) occurs when the radius of the 
ice-sheet is about one hundred miles. Once it extends beyond 
this limit, it will grow rapidly until it attains a radius of more 
than 500 miles, and then more slowly to a much greater radius. 


Conversely, during a period of amelioration of climate, a 
large ice-sheet will decrease slowly in size until it attains a 
radius of rather less than ten degrees (600 miles), after which 
it will shrink rapidly, even without any further change *of 
climate, until its radius is less than one hundred miles. This 
is exactly what happened in Scandinavia during the closing 
stages of the Quaternary Ice-Age. The ice-sheet withdrew 
slowly at first, until the distance from the centre to the southern 
edge was about nine degrees ; after this the retreat became 
more and more rapid until the ice-sheet had sunk to in- 
considerable dimensions. Here again the concordance of the 
theory with the observations is as good as could be wished. 

From this discussion we can see that if over an open polar 
ocean the winter temperature at the pole falls only as much 
as 0-6 F. below the freezing point of sea water, an ice-cap 
will develop, which will extend rapidly until it reaches a 
latitude of about 78. From this stage it will grow more 
slowly to about 65, unless in the meantime its growth is 
arrested by land, by the seasonal change of temperature, or 
by a warm ocean current. The ultimate lowering of the 
winter temperature brought about by the initial small fall of 
temperature of 0-6 F. will amount to about 45 F. 

This estimate of the cooling effect may appear large, 
but consider the difference of conditions a little way below 
and a little way above the critical point. Below the critical 
point all the cooled water remains on the surface in the form 
of ice, spreading freezing temperatures all round it, and even 
when the ice melts, the cold thaw water, owing to its com- 
parative freshness, is lighter than the warmer more saline 
water and remains on the surface, freezing again very readily. 
Fresh snow falls on the ice and increases its thickness, until the 
surface of the central parts of the Arctic Ocean resembles that 
of the Antarctic Ice Barrier, with correspondingly low tem- 
perature. On the other hand, above the critical point, the 
water as it cools becomes denser and sinks to the bottom, 
where it drains away as a ground current, to be replaced by 
warm surface currents from lower latitudes. Every inlet is 
flooded with warm water, no ice can form, and the temperature 
never sinks below the freezing point of sea water. The differ- 
ence between these two pictures shows that they may easily 
have a temperature difference of 45 F, 


This good agreement between the conditions found in the 
Arctic Ocean and the conditions deduced from theory suggests 
that the " non-glacial " winter temperature of the Arctic 
Oean may not be far below the freezing point of sea water, 
and that if the Arctic ice could once be swept away, it might 
find some difficulty in re-establishing itself. It therefore 
becomes an interesting speculation to try to determine what 
the " non-glacial " temperature of the Arctic would be. The 
problem may be stated thus : to determine what would be 
the distribution of temperature if sea water, with all its other 
properties unchanged, reached its freezing point and point 
of maximum density at a very much lower temperature than 
28 F. This problem may be attacked in two ways. One 
solution, which is due to F. Kerner (2), starts with an attempt 
to determine the temperature distribution over an entirely 
open ocean, and then superposes on this the effects of the 
present land and sea distribution. Kerner considers that 
there are only two regions in which the oceans are free from 
the effects of floating ice, namely, the seas of the East Indies 
and the centre of the North Pacific anticyclonic area, and 
he takes the temperatures of the sea surface in these two regions 
as the real oceanic temperatures which would obtain if there 
were no transference of heat across the parallels of latitude by 
ocean currents. From these temperatures, taking account 
of the influence of solar and terrestrial radiation, he calculates 
by various methods the " akryogenous " (i.e., " non-glacial ") 
temperature of every tenth parallel of latitude, and obtains 
for the North Pole an average annual temperature of 28 F. 
He computes the annual range of temperature as i3'5F., 
and this gives him for the January temperature 2i'3F., 
which he regards as the lowest limit for the temperature of an 
open ocean near the pole in the absence of ocean currents 
crossing the lines of latitude. (This, of course, is well below the 
freezing point of sea water, but we have guarded against this 
contretemps by a suppositions change of the freezing point.) 

Kerner next analyses the present distribution of temperature 
in January along the 75th parallel of latitude, and obtains a 
geographical formula which expresses the temperature of any 
point in this latitude in terms of the land and sea distribution. 
The geographical effects are twofold ; an elevation of tempera- 
ture by the Gulf Stream Drift, and* a depression of temperature 


owing to the cooling effect in winter of the large land-masses 
which nearly surround the Arctic Ocean in about 70 N. 
From Kerner's discussion it appears probable that the Gulf 
Stream effect is slightly the larger, and that the " non-glaci51 " 
temperature near the pole in an Arctic ocean with the present 
configuration would be a degree or two above the figure of 
21 '3 F. calculated for the open ocean. 

The second way in which a rough calculation of the " non- 
glacial " temperature of the North Pole may be made depends 
on my investigations of " continentality and temperature " 
described in Chapter VIII. , and referred to earlier in this 
chapter. By comparing the distribution of temperature with 
the distribution of land and of ice, I obtained measures of the 
thermal effects of land and of ice, in January and in July in 
different latitudes. These enabled me to eliminate the land 
and ice effects, but not the effect of ocean currents crossing 
the parallels of latitude, and consequently gave me the 
temperature of an open ocean under " non-glacial " conditions, 
with a steady interchange of water between equator and pole. 
The calculation was carried up to 70 N., but the figures 
calculated for January are very erratic. The results for July 
are more regular, since in that month the interference by ice 
is much less than in January, and the July figures gave : 

Latitude, N 40 50 60 70 

Mean July temperature, F. 64-6 55 -o 45-9 43*3 

By extrapolation the July water temperature of 90 N. would 
be about 40 F., and since there is very little land within 20 
of the North Pole, we may take the figure of 40 F. to represent 
the " non-glacial " July temperature of the North Pole with 
the present land and sea distribution. The annual range 
over an open ice-free polar ocean being 13*5 F. according 
to Kerner's calculation, the January " non-glacial " tempera- 
ture would be about 26*5F. There are, however, many 
objections to this method of calculation, and it cannot give 
us more than a rough approximation. 

Taking account of all three sources of information, the 
ice-conditions in the Arctic which suggest a January " non- 
glacial " temperature of not more than 24 F. (p. 39), 
Kerner's calculation of not less than 21-3 F., and my own 
rough calculation of 26'5*F., we find that the first of these 


figures, 24 F., is probably not far from the truth. This gives 
us some very interesting information about the stability of 
the present climatic regime. Quite a small rise in the general 
temperature of the earth, say, 2 F., would suffice to make the 
whole ice-sheet unstable in summer. This does not necessarily 
mean that it would completely break up and disappear during 
a single summer. The Palaeocrystic ice is so thick and 
extensive that in order to melt it a very large amount of heat 
would be required. Hence the most that could happen 
during a single abnormally warm summer would be that, 
after the fringe of thin ice of one winter's growth had been 
broken up, a border a few miles wide round the Palaeocrystic 
ice might be attacked before the winter cold came again to 
reverse the process. But suppose that there occurred a real 
and permanent change in the general climatic conditions, 
so that the warm summers came year after year, alternating 
with mild winters in which the outer fringe of ice did not form 
again with quite its old thickness or extent. Under these 
conditions, aided by the gradual drift of the whole ice-mass 
across the Arctic Ocean, the Palaeocrystic ice would gradually 
break up and be replaced by much thinner and looser ice. 
After this stage had been reached it might be possible for the 
ice to disappear completely during favourable summers, 
forming again in the following winter. 

Such a " semi-glacial " condition would require a nice 
adjustment of temperatures, for it seems that it would be 
possible only if the temperature in winter fell so little below 
the critical figure of 27'5F. that the initial stages in the 
formation of the ice-cap took place very slowly. If the worst 
of the winter were over by the time the ice-cap had attained its 
critical radius of one degree, the subsequent extension beyond 
this radius would be small and weak, and the ice-cap would 
be readily dissipated in summer. It will be shown in Part III. 
that this condition in which ice formed in winter but dis- 
appeared in summer, probably existed during the fifth to 
seventh centuries of the Christian era. 

If the general warming up went a little farther, so that the 
" non-glacial " winter temperature at the pole rose above 
the freezing point of sea water, the process would at first be 
very similar, though presumably more rapid. The Palaeo- 
crystic ice would melt back farther and farther each summer, 



but so long as it persisted with a radius exceeding one degree, 
an extensive ice-cap would form again in the winter. Finally, 
however, there would come a summer in which the ice-cap 
completely disappeared, and in the following winter it wo&ld 
not form again ; the climate would become definitely 
" non-glacial." 

Apart from the transitional or " semi-glacial " stage, 
which can have occurred but rarely in geological times and 
then only for short intervals, only two types of oceanic climate 


| 30 




90 60 70 60 

Latitude, Degrees 


Fig. 2. Temperature difference between " non-glacial " 
and " glacial " climate. 

are possible, the " glacial " and the " non-glacial/ 3 and these 
are very wide apart. The former is characterised by a very 
rapid decrease of temperature in high latitudes, the latter 
by a slow decrease. In order to show this, I have constructed 
Fig. 2, which gives the variations of temperature between 
latitudes 50 and 90. The straight line represents conditions 
in winter over an open ocean in which the temperature falls 
at the rate of o 9 F. for every degree of latitude, the water 
at the pole being one degree above the freezing point. The 
curve below it represents the temperatures over the same 


ocean in winter which would be the final result of a general 
initial fall of temperature by 5 F., causing the formation of 
an ice-cap in the way described above. The cooling power 
of ice is taken as 0-45 F. for each one per cent, of a ten-degree 
circle. The original fall of temperature by 5 F. has increased 
to 50 F. in high latitudes as the result of the formation of 
the ice-cap. 

This result greatly simplifies the problem presented by 
warm polar climates. Instead of having to account for 
changes of temperature of the order of 50 F. we have only 
to account for initial changes of 5 F. or so, since we can 
safely leave the floating ice to make up the odd 45 F. Before 
we are in a position to begin a serious search for the causes 
of these greatly reduced changes, however, it will be necessary 
to carry our preliminary studies a little further, and obtain 
some information about the winds and ocean currents, and the 
way they may have been modified by the change from a 
" glacial " to a " non-glacial " climate. 


(1) BROOKS, C. E. P. " The problem of mild polar climates." London, 

Q,. jf. /?. Meteor. Soc., 51, 1925, p. 83. 

(2) KERNER, F. " Das akryogene Seeklima und seine Bedeutung fur geolo- 

gischen Probleme der Arktis." Wien, Sitzungsber. Akad. Wiss., 131, 
1922, p. 153. 



THE weather at any place on any day is mainly 
governed by the winds which are blowing at the 
time ; similarly, the climate is largely determined by 
the winds which blow most frequently. The winds in one 
place are closely related to the winds at other places, the 
whole forming a more or less orderly system which ultimately 
depends on the differences of temperature between different 
latitudes. This system of winds over the earth as a whole is 
known as the circulation of the earth's atmosphere. Since 
the winds are bound up with the distribution of pressure, their 
discussion must also include the distribution of pressure in 
the various seasons. 

Figs. 3 and 4 show the average pressure distribution, reduced 
to mean sea-level, prevailing at present in January and July 
respectively. In January there is a belt of relatively low 
pressure (below 1,012 millibars) extending the whole length 
of the equatorial regions, with local minima over the 
continents south of the equator (South Africa, South America, 
and Australasia). On either side of this low-pressure belt are 
the sub-tropical high-pressure belts between 20 and 40 
latitude. The high-pressure belt in the Northern Hemisphere 
is especially well developed ; in addition to the maxima 
over the oceans there are marked anticyclones near the 
centres of the continents. In the Southern Hemisphere 
the anticyclones lie entirely over the oceans, with their centres 
rather nearer the eastern than the western shores. 

On the poleward sides of these high-pressure belts lie 
marked areas of low pressure. In the Northern Hemisphere 
these are in the form of isolated depressions over the oceans. 
One area, known as the Aleutian low, is centred near the 
Aleutian Islands in latitude 52 N. ; the other, the Icelandic 
low, lies west-south-west of Iceland in 60 N., and extends in 
a long tongue towards the Arctic Circle between Norway and 




4 8 



Spitsbergen. In its centre the average pressure is below 
996 mb. In the Southern Hemisphere, where the Southern 
Ocean is not broken up by land-masses, the area of minimum 
pressure is remarkably deep, and extends in a continuous 
belt completely round the globe in about 60 S. ; it is, 
however, accentuated in the Ross Sea and Weddell Sea. On 
the poleward sides of these minima, pressure rises again, 
especially towards the Antarctic continent ; in the Arctic 
Ocean the polar increase is masked by the Siberian and 
American anticyclones. 

In the chart for July (Fig. 4) the relative distribution over 
the two hemispheres is partly reversed. The equatorial 
low-pressure belt is still shown, but the lowest pressure is now 
found north of the equator, near Jacobabad in North-western 
India, about latitude 30 N. In the Northern Hemisphere 
the sub-tropical high-pressure belt is limited to two areas over 
the Atlantic and Pacific Oceans, but in the Southern Hemi- 
sphere it is nearly continuous in about 25 S. The Icelandic 
minimum still persists with greatly reduced intensity, but the 
Aleutian low has disappeared ; on the other hand, the low- 
pressure belt in the Southern Ocean is very sharply defined. 
As in January, pressure rises again near the poles. 

If the earth were not rotating about its axis, air would 
simply flow directly from areas of high pressure to areas of 
low pressure. Owing to the rotation the air is deflected, to 
the right in the Northern Hemisphere and to the left in the 
Southern Hemisphere, and in the free air it blows parallel 
with the isobars. Close to the earth's surface, however, the 
wind blows obliquely inwards towards the low pressure, and 
in the Northern Hemisphere a system of straight isobars with 
low pressure to the north and high pressure to the south gives 
surface winds from between west-south-west and south-west. 
Thus in accordance with the pressure distribution shown in 
the figures, on the eastern and south-eastern sides of the 
sub-tropical high-pressure centres there are great systems 
of winds, blowing towards the equator, known as the Trade 
Winds, north-east in the Northern and south-east in the 
Southern Hemisphere. Near the equator these two wind 
systems unite in a slow drift which on the whole is from east 
to west, though it is diversified by frequent calms and variable 
winds ; this region is known as the Doldrums. On the 


poleward side of the high-pressure belts occur westerly winds, 
which blow round the globe in temperate latitudes and are 
varied by frequent " cyclonic depressions." Finally, on the 
edge of the polar high-pressure areas easterly winds agin 
increase in frequency. 

This generalised system is modified by the continents, 
which tend to be occupied by high-pressure areas in winter 
and by low-pressure areas in summer. These are associated 
with continental systems of winds known as monsoons, blowing 
outwards from the continents in winter and inwards towards 
the continents in summer. The classic example of a monsoon 
area is Asia ; the winter pressure over Siberia, when corrected 
to sea-level, is the highest known on the earth, on the average 
exceeding 1,040 mb. south of Lake Baikal, while the summer 
low pressure in North-western India rivals in intensity the 
great barometric minimum in the Southern Ocean. There is 
a correspondingly great reversal of wind direction, for instance 
on the China coast the winds blow from the north or north-east 
with remarkable persistence for several months in winter, and 
from the south with almost equal steadiness in summer. The 
neighbouring continent of Australia south of the equator is 
undergoing the same alternation, though not so marked, in 
the reverse sense, so that when pressure is high in Asia it is low 
in Australia and vice versa ; consequently there takes place 
an immense ebb and flow of air between these two regions. 

It is evident that we have to deal with two factors which 
combine to form the main features of the distribution of 
pressure and winds over the globe. The first is known as the 
planetary circulation, the second is the influence of the 
continents and oceans and also of the distribution of ice ; it 
must have varied greatly from one geological epoch to another 
in accordance with the varying land and sea distribution, and 
may be termed the geographical circulation. The planetary 
circulation also cannot be regarded as a constant ; it is 
governed by the contrast of temperature between low and high 
latitudes, and before we can estimate the part which the 
atmospheric circulation has played in geological changes of 
climate, it will be necessary to discuss the planetary circulation 
somewhat more fully, especially in relation to the vertical 
and horizontal distribution of temperatures. 

Near the surface of the earth, temperature decreases upwards 


at the average rate of about 19 F. per mile, but this decrease 
does not continue indefinitely. It is found that at a height of 
several miles temperature ceases to fall, and above that height 
ifremains constant or even rises a little. The lower part of 
the atmosphere, in which temperature decreases upwards, is 
termed the troposphere, the succeeding layer in which there 
is no change with height is termed the stratosphere, the 
junction between them being the " tropopause." Now the 
tropopause is not always found at the same height ; it is 
slightly higher in summer than in winter, it is higher over 
anticyclones than over cyclones, and it is very much higher 
near the equator than near the poles. Over the equator its 
average height is about 18 kilometres (n miles), over the 
British Isles about n km. (7 miles), near the North Pole 
only about 7^ km. (less than 5 miles). This curious structure 
of the atmosphere has some very important consequences. 
In the troposphere the temperature at any given height is 
roughly proportional to that at sea-level ; the air five miles 
above the equator is warmer than the air five miles above the 
pole. But at a height of ten miles the conditions are very 
different. Let us work it out. At sea-level the temperatures 
are, say, 80 F. at the equator and 20 F. near the North Pole. 
Allowing a decrease of 19 F. for each mile of height, at five 
miles we have: equator, 15 F., pole, 75 F. But as 
we go on to a height of ten miles the temperature above the 
equator goes on falling, while that above the pole stays at 
75 F., and at ten miles we have: equator, noF., 
pole, 75 F., that is, the air ten miles above the pole is 
35 F. warmer than the air at the same height above the 
equator. In fact, the lowest temperatures naturally existing 
anywhere on earth are found at a height of about 10^ miles 
above the equator. Cold air is heavier than warm air, and 
therefore the colder the air above any locality, the greater will 
be the pressure there. 

The barometric pressure on any part of the earth's surface 
is the result of the whole column of air above it, but we may 
follow Sir Napier Shaw (i) in dividing the atmosphere into 
two parts, one above the level of 8 kilometres (5 miles) and the 
other below that level, and we may further suppose that at a 
height of about 20 kilometres (12^ miles) the pressure 
differences between different parts of the earth are small. The 


pressure at eight kilometres is inversely proportional to the 
temperature (on the absolute scale) of the higher layers of the 
atmosphere, and since in the upper air temperature increases 
from low to high latitudes, at eight kilometres the pressifre 
decreases from low to high latitudes. If the surface of the 
globe were fairly homogeneous this would result in the forma- 
tion of two great systems of westerly winds in the upper air 
(the " polar whirls ") blowing completely round the earth 
with their centres near the poles. Such a polar whirl may in 
fact exist in the Southern Hemisphere, but in the Northern 
Hemisphere, with its geographical and meteorological com- 
plexities, the circulation is greatly distorted, so that it is more 
correct to speak of a " zone of sub-polar whirls." This 
condition, however, must be an exception from the geological 
point of view, and during the warm geological periods the 
polar whirls were probably in existence in the upper air in 
both hemispheres. The reason for the exception in the 
Northern Hemisphere at present is not the low temperature 
over the Greenland ice sheet, since the layer of abnormally cold 
air above the ice is comparatively thin ; the circulation 
apparently takes place round the low-pressure centre and is 
due to the fact referred to previously, that, above the low 
pressure the stratosphere extends down to a low level and is 
therefore abnormally warm. 

In the stratum between the surface and eight kilometres 
the average temperature decreases from lower to higher 
latitudes, and the pressure at the surface due to this layer of 
air alone is therefore greater in high than in low latitudes. 
The effect of this layer would be to cause easterly winds at 
the surface in all latitudes. Thus we have two apposing 
tendencies ; the temperature distribution above eight kilo- 
metres tends to produce westerly winds at all levels from the 
surface up to nearly twenty kilometres, while the temperature 
distribution below eight kilometres tends to produce easterly 
winds at the surface. Which of these two directions, ea 4 st or 
west, predominates in the resultant surface wind of any 
latitude, depends on the vertical temperature distribution. 
If the decrease of temperature towards the poles is rapid, 
as in the " glacial " periods, the weight of the cold air in high 
latitudes in the lowest eight kilometres may be more than 
sufficient to counterbalance the pressure difference between 


the equator and the poles at eight kilometres, and there will 
be a polar cap of east winds as at present. But if the decrease 
of temperature towards the poles is slow, as it was during the 
" non-glacial " periods, the weight of the lowest eight kilo- 
metres of air may not be enough to counterbalance the pressure 
difference between the equator and the poles at eight kilometres, 
and in that event westerly winds will prevail at the surface 
up to the immediate neighbourhood of the poles. This must 
not be interpreted to mean that there were no depressions, 
or only one stationary depression concentric with the pole. 
Some theoretical work of H. Jeffreys (2) leads him to the 
conclusion that the friction of the wind against the earth's 
surface must inevitably introduce a system of moving cyclones 
surrounding the pole. In the absence of glacial anticyclones, 
however, it appears that cyclones would be fewer and less 
intense than at present, and would occur as a rule in very 
high latitudes. 

At this point we may pause to consider a possible objection. 
Since the atmospheric circulation depends on differences of 
temperature, it would seem to follow that the greater the 
temperature difference between equator and poles, the stronger 
would be the atmospheric circulation, and consequently the 
greater the amount of heat carried from low to high latitudes 
by the winds and wind-driven ocean currents. This would 
tend to restore the balance and keep the temperature difference 
between equator and poles more or less constant. This way 
of looking at the atmospheric circulation is very plausible, but 
it ignores the fact brought out in the preceding discussion that 
there is a critical point for the planetary atmospheric circulation 
just as for the distribution of temperature in a polar sea. 
In the glacial state, with low temperatures near the poles 
the excess of air density in the layer of the atmosphere below 
eight kilometres causes easterly winds in high latitudes ; 
these have an equatorward component at the surface which 
hinders the poleward surface component of the westerly winds 
from carrying warm air into high latitudes. This shutting 
out of the warmth-bringing equatorial air helps to keep the 
polar regions cold. In the " non-glacial " state, the excess 
of density in the lowest eight kilometres is not sufficient to 
counterbalance the warmer stratosphere of high latitudes and 
the winds are therefore westerly with a poleward surface 


component, up to high latitudes. The influx of warm 
equatorial air then helps to maintain the high polar tem- 
peratures. Owing to the existence of this critical point tke 
atmospheric circulation, so far from smoothing out the tem- 
perature contrast, maintains or even magnifies it. 

The boundary between the polar east winds and the 
temperate west winds (the " polar front ") is at present 
the chief seat of the development of the barometric 
depressions of temperate latitudes. There seems to be no 
doubt that the presence of two adjacent air currents at 
different temperatures which are in motion relative to one 
another is an important factor in bringing these depressions 
into existence (3). Owing to the frequent passage of depres- 
sions the average pressure in about 60-65 latitude is lower 
than it would otherwise be. These belts of cyclonic activity 
and low pressure limit the poleward extension of the sub- 
tropical anticyclones and help to cause the relatively sharp 
differentiation of the climatic zones at present ; in the absence 
of polar east winds we should expect a gradual fall of pressure 
from maxima in middle latitudes to minima near the poles. 
The chief centres of storminess would be found over the oceans 
in the Arctic and Antarctic regions ; in middle latitudes there 
would be only occasional feeble depressions moving slowly 
along irregular tracks. 

The zone of westerly winds which blow round the world 
in middle latitudes is broken up to some extent near the 
surface by travelling depressions and anticyclones, but at a 
height of a few miles the winds are much stronger and more 
steady. A steady west wind blowing round the globe would, 
however, be unstable; as described for example by C. G. 
Rossby (4), any slight disturbance of the west-east movement 
would set up wave-like disturbances, with the crests towards 
the pole and the troughs towards the equator. The wave- 
length depends on the strength of the winds ; the average 
wind velocity in winter in the northern hemisphere at present 
is such that the wave-length is about 3,000 miles, and this is 
the approximate distance between the Aleutian and Icelandic 
low-pressure centres (the corresponding centre in about 60 E. 
is masked by the Siberian winter anticyclone). The weaker 
the winds the shorter the wave-length ; hence when for any 
reason the average speed of the westerly winds falls off, the 


circulation tends to break up into a number of smaller cells. 
The deep single Aleutian and Icelandic low pressure centres are 
replaced by smaller and less stable double centres. The winds 
are less strong, but the exchange of air between high and low 
latitudes is maintained by larger north-south and south-north 

Since the strength of the west-east circulation depends on 
the temperature gradient between low and high latitudes, in 
the non-glacial periods we should expect a weak circulation 
and hence numerous small areas of low pressure. This 
would tend to equalise climatic conditions along the same 
parallel of latitude, whereas the large semi-permanent Aleutian 
and Icelandic lows which predominate at present lead to 
great extremes, e.g. between Labrador and north-west Europe. 

In the early stages of the Quaternary glaciation, as pointed 
out by R. F. Flint and H. G. Dorsey (6), the zonal circulation 
must have increased in strength. At present, when the 
circulation is especially strong the Icelandic low tends to spread 
eastwards ; a similar change in the Quaternary would have 
facilitated the eastward extension of the Scandinavian 
glaciation. On the other hand strong west winds crossing 
the Rocky Mountains would give Fohn effects on the leeward 
side and so limit the eastward extension of the Gordilleran 
ice-sheets, favouring a separate centre of glaciation in eastern 
Quebec and Labrador. When fair-sized ice-sheets had 
developed in the north, however, semi-permanent glacial 
anticyclones would form which would displace the tracks of 
depressions southward. This would weaken the zonal circula- 
tion ; depressions would tend to stagnate south of the ice- 
sheets, facilitating the extension of the latter southward on 
their western margins while farther east comparatively warm 
southerly winds checked the southward expansion. These 
ideas are interesting and warrant further investigation, but the 
pattern suggested by Flint and Dorsey is complicated by 
other factors such as isostatic changes due to the weight of 
the ice. 

In tropical and sub-tropical regions the existing planetary 
circulation over the oceans (equatorial low-pressure belt, trade 
winds, sub-tropical anticyclones) is clearly the result of zonal 
temperature differences and the rotation of the earth. In 
warm periods thermal zones must have existed though probably 


somewhat less marked than now, and there is no reason to 
suppose that trade winds and sub-tropical anticyclones did not 
exist. The absence of the polar fronts, however, and the pole- 
ward displacement of the tracks of depressions would permit 
the anticyclones to extend farther poleward, just as they do 
now in the North Atlantic in summer, when the south-north 
thermal gradient is smallest, compared with winter, when the 
thermal gradient is greatest. 

Fortunately Nature herself carries out for our inspection an 
experiment which suggests the way in which the general 
atmospheric circulation to be expected during periods with a 
small temperature difference between equator and poles would 
differ from the circulation during periods with a large temper- 
ature difference. In summer the air over the Arctic Ocean 
is now more than 30 F. warmer than in winter, while in 
equatorial regions there is very little change of temperature 
throughout the year. The temperature difference between 
equatorial and polar regions is therefore about 50 F. in 
summer and more than 80 F. in winter. As a rough approxi- 
mation we may say that the temperature difference between 
low and high latitudes during the present summer in the 
Northern Hemisphere is similar to the temperature difference 
which existed during winter in the warm periods. Nature 
has a habit of complicating her experiments with irrelevant 
details, and in all the oceans but one there are, even in summer, 
large quantities of ice and ice-cooled water which, coming into 
contact with warm ocean currents, produce great differences 
of temperature in short distances, but in the North Pacific 
Ocean this complication is reduced to a minimum, and we may 
illustrate the probable system of winds during the winter of a 
warm period by the system of winds prevailing at present over 
the North Pacific in summer (Fig. 6). Of course it is not to be 
expected that this will give us an exact picture of what happened 
during a warm period ; the differences in other oceans, 
and especially in the Southern Hemisphere, would mevit^bly 
introduce some modifications, but at least it will serve as a 

The distribution of pressure at present has been shown in 
Figs. 3 and 4. In January (Fig. 3) the anticyclone is small 
and sharply defined ; it does not extend north of 40 N., and 
between 40 N. and 60 N. lies a well-marked area of low 


pressure, the Aleutian low. In July the anticyclone is less 
sharply defined, but extends almost to 60 N., and the Aleutian 
low has completely disappeared. This disappearance of the 
cyclonic centre from the North Pacific in summer is of great 
importance ; the Icelandic minimum in the North Atlantic 
does not disappear during summer and the low-pressure belt 
in the Southern Hemisphere also persists throughout the year. 
There seems no doubt that this peculiarity of the North Pacific 
is directly due to the absence of ice in summer both floating 
ice in the sea and large ice-sheets on the neighbouring land- 
masses. The surface winds over the Pacific corresponding 
with the pressure distribution in January and July are shown 

' < / /.. f. f f f f f^ r .-f f ^-r- 


Fig. 5. Winds over North Pacific, January. 

in Figs. 5 and 6. In January (Fig. 5), representing winter 
conditions at present, a belt of north-easterly trade winds 
extends completely across the ocean south of latitude 25 N. 
and is almost entirely separated by a belt of calms and variable 
winds from another system of winds, mainly westerly, between 
30 and 50 N. North of 50 N. the prevailing wind is from 
north-east. Thus the system of winds in January, while it is 
calculated to bring mild winters to the American coast south 
of 55 N., greatly intensifies the rigour of the winter in sub- 
polar latitudes and on the eastern coasts of Asia. 

In July (Fig. 6), representing summer conditions at present 
and winter conditions during the warm periods, the system 
of winds is very different. The direction of the north-east 


trade is more easterly, and west of 1 80 longitude even south- 
easterly ; it passes without a break into a great system of 
southerly winds which blow from 30 N. to the Arctic Circle. 
The south-easterly direction of these winds north of about 
latitude 50 N. is due to the monsoonal inflow into the great 
land-mass of Asia ; during the warm periods the winds 
probably retained their south-westerly direction into high 
latitudes. These winds bring high temperatures over the 
whole of the North Pacific coasts (except the western coast 
of the United States), and are especially favourable to Alaska 
and North-eastern Asia, which enjoy a warm climate at this 



160 E 180 W 160 


Fig. 6. Winds over North Pacific, July. 

season. The winds induce favourable oceanic currents, and 
in spite of the narrowness and small depth of Bering Strait 
some warm water succeeds in penetrating into the Arctic 
Ocean through that opening in summer. The Arctic Ocean 
retains a large amount of floating ice throughout the summer, 
so that the polar cyclone cannot develop properly, and the 
southerly winds of the North Pacific are greatly weakened ; 
there seems no doubt that if the Arctic Ocean were free of 
ice the system of southerly or south-westerly winds would attain 
great steadiness. If at the same time the Bering Strait were 
replaced by a wide and deep gap, practically the whole surface 
of the ocean north of 30 N. would be set in motion in a north- 
easterly direction, and an immense volume of warm water 
would be driven into the Arctic Ocean. 


This picture of the pressure and winds during the geological 
periods characterised by widespread warmth, which we have 
obtained first from some theoretical considerations, and 
secondly from an examination of present summer conditions 
over the North Pacific, fits in admirably with what we know 
of the climates of these warm periods. The system of stable 
southerly winds, extending across the middle latitudes, would 
give them fine quiet weather in place of the present succession 
of storms. As we saw in the Introduction, apart from the 
warmth and rich vegetation of the polar regions, the most 
striking feature of these periods was the widespread develop- 
ment of semi-desert conditions in temperate regions. True 
deserts were probably less extensive than they are to-day, 
owing to the smoothing out of the zonal contrasts, but very 
large regions had a " Mediterranean " type of climate, with a 
small rainfall during the mild winter and a long dry hot 
summer. This explains, for example, the widespread dis- 
tribution of the characteristic " sub-tropical " (i.e., Medi- 
terranean) vegetation of the first half of the Tertiary period. 

Let us look at the reverse of this picture, and see what 
would happen during an ice-age. An extension of the cold 
areas over a large part of the present temperate regions, such 
as occurred during the Quaternary, would bring the " polar 
front " between the polar east winds and the temperate west 
winds nearer to the equator, and by increasing the temperature 
contrast between low and high latitudes would increase the 
storminess. This means that the sub-tropical high-pressure 
belts would be sharply limited on their poleward sides. The 
result would probably be an intensification of the present 
winter conditions a small but intense anticyclonic belt in 
about 20 to 25 latitude, a narrow belt of powerful trade 
winds, and a deepened equatorial trough of low pressure. 
The circulation would be much less stable than that of the 
warm periods, and the anticyclonic belts would be subject to 
grea* and rapid displacements. Hence the rainfall would be 
increased over all the tropical and sub-tropical regions ; 
outside the new storm tracks the increase would be greatest and 
most regular near the equator, while towards the tropics the 
rainfall would be less in amount and very variable from year 
to year. Here again the theoretical conclusions are supported 
by geological results ; outside the great ice-sheets there is 


evidence of a much greater rainfall (snowfall on the mountains 
in two belts, one along the new storm tracks a short distance 
equatorward of the ice-edges, and the other along the equator. 
The former gave rise to the great lakes of the Great Basin of 
America and of the interior of Asia, and to a large number of 
mountain glaciers, the latter to the greatly increased lakes of 
Central Africa and to the glaciers of Kenya, Kilimanjaro, 
Ruwenzori, and parts of the Andes. Between these two belts 
the evidence of greater precipitation is more indefinite and 

We must now return to the geographical circulation. 
This we have seen is characterised by the presence over the 
continents of high pressure in winter and low pressure in 
summer, resulting in monsoonal winds. The intensity of 
the monsoons over a continent depends on a number of 
factors the limits of latitude, the size, the presence of large 
arid basins surrounded by mountain ranges, and the strength 
and direction of the winds over the neighbouring seas. The 
intensity of the Siberian winter anticyclone is due partly to 
the great size of Eurasia, and very largely to the way in which 
its surface is broken up by mountain ranges. The highest 
pressure occurs over the great enclosed basin south of Lake 
Baikal, from which the cold air finds difficulty in escaping. 
The winter cooling is of course essential to the development 
of the anticyclone, but the existence of the anticyclone in turn 
intensifies the cold. In North America, where owing to the 
absence of transverse mountain ranges the cold air finds less 
difficulty in escaping, the winter anticyclone is comparatively 
feeble. A good example of the effect of favourable orographical 
conditions on the pressure distribution is the Iberian Peninsula, 
which is occupied by a well-marked anticyclone in winter. 
In North-west Europe, on the other hand, the influence of the 
Icelandic minimum and the barometric depressions which 
originate in the Atlantic make the maintenance of a winter 
anticyclone very difficult. 

The establishment of continental low pressure in summer 
depends on similar factors, but the centres occur nearer the 
equator than do those of the winter anticyclones. In Asia the 
area of low pressure in July extends over a large part of the 
continent, but the actual minimum occurs comparatively near 
the sea in North-west India, and is very intense ; this position 


of the minimum is due largely to the position of the mountain 
ranges which interfere with the free circulation of the air (6) 
and entirely inhibit the North-east Trade, which is the natural 
wind of that latitude. Directly to the westward over the 
Sahara, where the air is as hot or hotter, pressure is much 
higher, and in fact throughout the year the Sahara is practically 
a continuation of the belt of the North-east Trades, probably 
because the mountain ranges are not high enough to shut out 
these winds entirely. 

These scattered instances show how difficult it is to analyse 
the geographical circulation exactly, but they do give us some 
basis for estimating the results of various geological changes on 
the local wind circulations. Consider, for instance, a warm 
period in which there are extensive oceans and some flat 
continents of moderate size, little diversified by mountain 
ranges. It seems that these continents would not greatly 
modify the planetary circulation described above. In summer, 
when the winds are generally weakest, the continents in low 
and middle latitudes would be hot, and there being no 
mountain ranges to cause the ascent of air, the only rainfall 
would occur in sporadic thunderstorms and squalls. In 
winter they would be cooler than the oocans, but in the absence 
of a polar reservoir of cold air, they would not be intensely 
cold, especially since the prevailing winds in temperate zones 
would be from the equator. The centres of the anticyclones, 
as now, would probably lie over the oceans, but there would 
be a tendency for the high-pressure areas to extend nearer the 
poles over the continents, giving southerly winds on the west 
coasts and westerly winds on the east coasts. A diagrammatic 
reconstruction of the pressure and winds during winter and 
summer in a warm period is shown in Fig. 7. 

There are two special parts of the geographical circulation 
which demand further notice. One of these is the south-west 
monsoon of the Asiatic continent ; the other is the circulation 
over ice-sheets. The south-west monsoon (6) is remarkable 
because it involves a large transference of air from one hemi- 
sphere to the other ; there is a continuous pressure gradient 
from the sub-tropical anticyclone in the South Indian Ocean 
to the minimum over Asia, and the surface winds are south- 
easterly south of the equator and south-westerly north of the 
equator. The air-flow actually in places surmounts the 



mighty barrier of the Himalayas, nowhere less than 12,000 feet 
in height, and arrives in Tibet as a dry descending current. 
Now it is possible that under certain conditions there may be 
a quite considerable difference between the mean annual 
temperatures of the two hemispheres, leading to a more or 
less permanent circulation of the type of the south-west 
monsoon. A circulation of this type affords a possible 


Fig. 7. Reconstruction of pressure and winds during 
a warm period. 

explanation of the peculiar climate of the Upper Carboniferous 
period (see Chapter XV.). 

The circulation over a large ice-sheet is of great importance 
for the study of the meteorology of glacial epochs. A surface 
of snow and ice reflects a large part of the solar radiation 
which falls on it, and, owing to the dryness of the air above it, 
also radiates freely and receives little return radiation from 
the air. Hence during most of the year the surface is very 
cold, and this cools the lower layer of the air in contact with it. 
Since cold air is relatively heavy, barometric pressure must be 
higher over a large ice-sheet than at the same level over 


neighbouring seas. This is interpreted by W. H, Hobbs (7) 
to imply that an ice-sheet is the site of a nearly permanent 
anticyclone, with outflowing winds on all sides, supplied by 
air which flows in at higher levels and descends over the 
ice-sheet. This raises the difficult question of the supply of 
moisture to maintain the flow of ice, for descending air is 
warmed by compression and is normally dry, but Hobbs pointed 
out that the surface of the ice is intensely cold, and may 
be much colder than the air at a height of several thousand 
feet. The moisture of the upper winds reaches ground level 
in the form of vapour, but a large part of it is immediately 
condensed as ice-mist or deposited directly as hoar-frost. The 
outflowing winds sweep these crystals before them and so 
maintain the marginal parts of the ice-sheet. 

There is undoubtedly a good deal of truth in this theory. 
Owing to difficulties in reduction to sea-level, pressure on 
the surface of a large ice-sheet cannot be compared directly 
with that over the surrounding ocean, but the winds do on the 
whole tend to blow outwards and must be made good by 
descending currents, while the vertical distribution of tempera- 
ture is such that there must be some condensation as ice- 
crystals or hoar-frost. On the other hand, the supposed 
high pressure may be entirely due to a quite thin layer of air 
near the ground, above which the structure ceases to bear any 
relation to an anticyclone, while the outflowing winds may be 
purely katabatic rivers of cold air such as flow down any 
slope on a cold clear night and have no relation to the general 
pressure distribution. In such a case the normal processes 
of precipitation could go on unchanged above the surface 
layers of air, the only difference being that precipitation would 
fall entirely as snow. It is very unlikely that the deposit of 
hoar-frost could suffice to supply the enormous quantities of 
ice which issue from an ice-sheet each year as glaciers, icebergs 
and glacial streams. 

F. . Matthes (8) discussed the light thrown on this question 
by observations at Eismitte near the centre of Greenland 
during the Wegener expeditions of 1929 and 1930-31, the 
latter covering more than a year of continuous instrumental 
readings. These observations show clearly that quiet fine 
conditions are the exception and stormy cloudy conditions 
the rule, that the katabatic winds are feeble and easily 


overpowered by storm winds, and that by far the greatestfactor 
in the nourishment of the ice-sheet is ordinary snow. The 
great oscillations of pressure and temperature closely resemble 
though much lower in the scale, those in typical cyclonic, 
regions such as New England. The winds are strong, but 
blow mainly from east, with a secondary maximum from 
S.S.E. True precipitation is actually more frequent than on 
the west coast of Greenland and not much less common than 
on the rain-swept coast of Norway, but owing to the low 
temperatures the total amount is not large, the average being 
estimated from the firn layers as 12-4 inches of water a year. 

The frequency of easterly winds is not due entirely or even 
mainly to katabatic winds from the ice divide to the east. It 
means that depressions pass mainly to the south, but extend 
their influence right to the centre of the ice-cap. There is 
now no difficulty about the supply of moisture, which is 
derived from the Atlantic Ocean and carried inland, over- 
riding the shallow katabatic winds of the coast, to be condensed 
into snow over the ice-shed and swept on to be gradually 
deposited on the leeward slopes. Atmospheric pressure is 
undoubtedly higher than it would be if the surface, at the same 
level, were unglaciated, but it appears that the sub-continent 
is not big enough to form the basis for a self-supporting 
glacial anticyclone. 

The Antarctic presents a more difficult problem. W. 
Meinardus (14) considered that the Antarctic anticyclone is 
limited to the lowest 2,000 metres, above which the circulation 
is cyclonic, and since the greater part of the Antarctic continent 
is above this level, he supposed that the greater part of the land 
is subjected to cyclonic air motion. Sir George Simpson 
pointed out (9) that an extensive ice-covered plateau must 
be occupied by a glacial anticyclone just as if it were at sea- 
level ; he accordingly divides the Antarctic continent into two 
parts, a plateau at about 3,000 metres (10,000 feet) and a 
plain near sea-level. The latter is occupied by an anti- 
cyclone at sea-level, but owing to the rapid decrease of pressure 
with height, conditions in the free air are cyclonic at a height 
of about 3,000 metres. Thus at the latter height there is an 
anticyclone at the surface of the plateau and a cyclone in the 
free air above the plain. 

It appears that an ice-sheet does not develop a stable 


glacial anticyclone until it attains a certain size. The 
Greenland ice-sheet is not broad enough to prevent the 
influence of large depressions from extending even to its 
centre, but these depressions are deflected southwards by 
the ice-sheet so that their centres rarely pass directly over 
the central regions of Greenland. The greater part of the 
Antarctic continent is immune from travelling depressions. 
On the other hand, the smaller ice-masses of Iceland and 
Spitsbergen appear to have little effect on the pressure 
distribution. Thus the critical point comes at a diameter 
somewhat larger than the width of Greenland. In 75 N. 
Greenland is about 650 miles across. From studies of the 
January temperature distribution over land areas in latitude 
50 to 70 N. (10), which are normally snow covered in winter 
and are therefore similar to ice-sheets in their effect, I think 
the diameter which a circular ice-sheet must reach before it 
begins to dominate the pressure distribution is between seven 
hundred and a thousand miles. When the diameter is less 
than 700 miles, the normal winds of the region sweep over 
the ice-sheet without much hindrance, and the only effect 
of the ice is to cool the air slightly by conduction. When the 
diameter reaches, say, 1,000 miles, a glacial anticyclone 
develops, with clear skies and intense cooling by radiation. 
The outwardly-directed winds spread Arctic conditions in a 
broad zone round the margin of the ice, and may even result 
in the " sympathetic " glaciation of a neighbouring mountain 
range. It is probable that the great development of Alpine 
glaciers during the Quaternary was partly due to cooling by 
the winds blowing off the Scandinavian ice-sheet. It will 
also be remembered that during the earlier stages in the final 
retreat of the Scandinavian ice-sheet the ground vacated by 
the ice was occupied by a dwarf flora of Arctic plants, but 
later, when the ice-sheet was smaller, the retreating edge was 
immediately followed by a temperate flora. This probably 
indicates the stage at which the glacial anticyclone broke 

Every cause or factor which is put forward to explain 
climatic changes has to take into account the modifications 
which would be introduced into the atmospheric circulation 
by its operation. Even the possible occurrence of modifications 
of the circulation sufficient by themselves to give rise to great 


climatic changes, without the intervention of any other 
factor, has been discussed. Thus W. H. Dines remarks (i i) : 
" There seems to be no particular reason why the winds known 
as the c trades ' should not be westerly and the winds of 
temperate latitudes easterly. Perhaps such a system is possible 
and might be stable if once established. It would explain 
the glaciation of North-western Europe, for it would very 
greatly lower the temperature of that region, but it is not 
feasible as an explanation of the glacial epoch, because it 
would raise the winter temperature of North America." 
A restoration of the meteorological conditions of the Quatern- 
ary Ice-Age was attempted by the late F. W. Harmer (12) 
on the assumption that the glaciations of North America 
alternated with those of Europe. It is hard to conceive of 
great changes such as these without some ulterior reason, 
such as a change in the land and sea distribution or in the 
solar radiation, and we must regard the part which the 
atmospheric circulation plays as that of a regulator, at times 
perhaps an amplifier, but probably not an originator of major 
climatic oscillations. 

It must be admitted, however, .that the part played by the 
circulation of the atmosphere in climatic changes is not yet 
fully understood. In the past hundred years, for example, 
there has been a marked recession of glaciers in all parts of 
the world, accompanied by a large rise of temperature in the 
Arctic and a rise of winter temperature over a much wider 
region. R. Scherhag (13) attributes these phenomena to a 
strengthening of the atmospheric circulation and consequently 
of the Gulf Stream, but this only pushes the problem one stage 
further back, i.e. 9 to the cause of the stronger circulation. 
The answer certainly does not lie in a change of land and sea 
distribution, and to the best of our knowledge there has been 
no appreciable change of solar radiation. It is not unlikely 
that the cause lies in the atmosphere itself, or in its inter- 
actions with the oceans, owing to some process initiated almost 
by " accident " in the constant turmoil of depressions and 
anticyclones, but which, once begun, will automatically 
increase in intensity until it becomes unstable or is reversed 
by some other " accident." The atmosphere and hydrosphere 
are so vast that such self-reinforcing actions may well persist 
for many decades. It is possible that the majority of temporary 


swings of climate are of this nature. If so, they cannot be 
said to have a " cause," any more than can a run of luck in a 
game of pure chance. 


(1) SHAW, SIR NAPIER. " The air and its ways." London, 1924. 

(2) JEFFREYS, H. " On the dynamics of geostrophic winds." London, Q. J* R 

Meteor. Soc., 52, 1926, p. 85. 

(3) BJERKNES, J., and H. SOLBERG. " Life-cycle of cyclones and the polar front 

theory of atmospheric circulation." Kristiania, Geofysiske PubL, 3, No. i, 

(4) ROSSBY, C. G. " The scientific basis of modern meteorology." Washington, 

Tearb. Agric., 1941, p. 599. 

(5) FLINT, R. F., and H. G. DORSEY. " lowan and Tazewell drifts and the North 

American ice-sheet." New Haven, Amer. J. Sci., 243, 1945, p. 627. 

(6) SIMPSON, G. G. " The south-west monsoon." London, (. J. R. Meteor. 

Soc., 47, 1921, p. 151. 

(7) HOBBS, W. H. " The glacial anticyclones. The poles of atmospheric 

circulation." New York (Macmillan), 1926. 

(8) MATTHES, F. E. " The glacial anticyclone theory examined in the light of 

recent meteorological data from Greenland." Trans. Amer. Geoph. Union, 
27, 1946, Pt. I, p. 324. 

(9) BRITISH ANTARCTIC EXPEDITION, 1910-1913. Meteorology, vol. i. Dis- 

cussion, by G. G. SIMPSON. Calcutta, 1919. 

(10) BROOKS, C. E. P. " Continentality and temperature." London, Q,. J. R. 

Meteor. Soc., 43, 1917, p. 164. 

(11) DINES, W. H. "Circulation and temperature of the atmosphere." 

Washington, D.C., Monthly Weather Review, 43, 1915, p. 551. 

(12) HARMER, F. W. " The influence of the winds upon climate during the 

Pleistocene epoch : a pabeometeorological explanation of some 
geological problems." London, Q,. J. Geol. Soc., 57, 1901, p. 405. 

(13) SCHERHAG, R. " Die Erwarmung der Arktis." Copenhague, J. Cons. 

int. Explor. Mer., 12, 1937, p. 263. See also Meteor. Mag., London, 73, 
'938, p. 29. 

(14) DEUTSGHE-SttDpoLAR-ExpEDiTiON, 1901-1903. Ill Bd., Meteorologie. 

Berlin, 1911. 



OWING to the high specific heat of water, the great 
oceanic currents and the variations in the surface 
temperature of the sea to which they give rise are of 
very great climatic importance. The classic example of this 
is the Gulf Stream Drift and the high winter temperatures of 
North-west Europe, but there have been still more notable 
instances in the geological past, in the highly favourable 
climates of the polar regions during the warm periods. It is 
estimated that at present about half the transfer of heat from 
low to high latitudes is due to ocean currents, the remaining 
half being due to interchange of air. Ocean currents are due 
chiefly to two causes, differences of density, and the winds, 
which drive before them the surface layers from which motion 
is imparted by friction to the underlying layers. If two 
masses of water of different densities lie side by side a circula- 
tion will be set up between them, resulting in a surface flow 
from the lighter to the heavier mass, and a flow at a greater 
depth from the heavier to the lighter, and if the earth were at 
rest this would continue until the horizontal differences of 
density had been removed and all the heavy water lay at the 
bottom, with all the light water on top. Owing to the earth's 
rotation, currents of water are deflected to the right in the 
Northern Hemisphere and to the left in the Southern Hemi- 
sphere just like currents of air, and the ultimate flow is at 
right angles to the gradient of density, giving two currents 
moving side by side in opposite directions. 

Density depends almost entirely on two factors, the 
temperature and the salinity. With falling temperature, 
water increases in density until it reaches a temperature of 
4 C. (39 F.) if it is fresh, but about -2 C. (28 F.) if it is 
average sea water. Since 28 F. is also the freezing point of 
sea water, the latter on being cooled will always sink through 
water of equal salinity, while fresh water at a temperature 



of 39 F. will sink through fresh water having either a 
higher or lower temperature. The temperature of the oceans 
almost everywhere falls with increasing depth, and the lower 
parts of the oceans are occupied by a stratum of water at about 
39 F. 

Density also increases with salinity ; the average salinity 
is greatest in the sub-tropical open oceans where evaporation 
is great and rainfall slight. Since these are also in general 
the areas where the temperature is high, the effect of salinity 
on density partly balances that of temperature. Density 
becomes especially great where an ocean current from the 
tropics penetrates into high latitudes, losing much of its heat 
but retaining a high salinity. Thus the relatively warm salt 
water which originates in the Gulf Stream and penetrates into 
the Arctic Ocean is heavier than the colder but much fresher 
water of local origin ; the latter remains on the surface and 
freezes readily, giving rise to great quantities of floating ice. 
Fresh water is always lighter than sea water of average salinity 
(35 P ai% ts per thousand) at temperatures which are met with 
in nature. 

In the oceanic circulation as it is developed at present 
(Fig. 8), the winds apparently play a much greater part than 
differences of density, especially in tropical and temperate 
latitudes, where the direction of the ocean currents is almost 
everywhere the same as that of the air currents. On a non- 
rotating globe this agreement would be easy to undertsand, 
but as it is, the matter is more complex. The wind drives the 
surface layers before it, the movement being communicated by 
friction and vertical interchange of water to the underlying 
layers, but owing to the rotation of the globe, the surface 
current caused by a steady wind is inclined to the wind 
direction at an angle of 45 degrees, to the right in the Northern 
Hemisphere and to the left in the Southern Hemisphere. As 
we go below the surface the currents deviate more and more 
from the winds, and the main mass of the water moves at right 
angles to the direction of the wind. Consider now the case 
of a centre of high pressure in the Northern Hemisphere round 
which the winds circulate in a clockwise direction. All these 
winds will be driving the water to the right, that is, towards 
the centre of the system. The result will be that water is 
piled up in the centre ; we shall have an oceanic " hill " 



down the slopes of which the surface water will commence to 
run. The rotation of the earth will deviate this water to the 
right, and when a steady state is reached it will be flowing 
round the central " hill " in a clockwise direction. The final 
result will be in fact as we now find it, a system of oceanic 
currents surrounding a central area of stagnant ocean, closely 
resembling the system of winds blowing round a central area 
of calm. The tendency of the winds to drive water towards 
the centre is just balanced by the tendency of the accumulation 
of water in the centre to flow outwards. Winds irregular in 
direction and velocity are less effective than steady winds in 
causing ocean currents, and with variable winds the angle 
between the resultant wind and the surface current is generally 
less than 45 degrees (i). 

The best known system of currents is that in the North 
Atlantic (2, 3), (Fig. 8). Commencing with the tropical 
part of the Atlantic Ocean, we find that the North-east and 
South-east Trades give rise to currents which turn more and 
more to the eastward and increase in volume as they approach 
the equator, until they unite to form the Equatorial Current. 
For convenience, the northern and southern portions of this 
current are called the North and South Equatorial Currents, 
but there is no definite dividing line except that from May or 
June to November a very shallow current, the Equatorial 
Counter Current, sets eastward where the winds are weakest 
between about 3 and 10 N., ultimately entering the Gulf 
of Guinea. The greater part of the North Equatorial Current 
turns north-westward as the Antilles Current, which passes 
between Cuba and the Bahamas, and unites with the Gulf 
Stream flowing through the Strait of Florida. The Antilles 
Current is estimated to convey nearly forty cubic miles of 
water an hour past Porto Rico. 

Owing to the greater strength of the South-east Trades, 
the South Equatorial Current is stronger and steadier than the 
North Equatorial. It is directed slightly north of west ; 
striking Cape San Roque on the Brazilian coast in 5 S., it 
divides into two branches, of which the southern turns south- 
westwards as the Brazilian Current, while the northern and 
more extensive passes along the coast of Guiana and unites 
with the western branch of the North Equatorial Current. 
The combined current flows towards the coasts of Honduras 


and Yucatan, and thence mainly through the Yucatan Channel 
into the Gulf of Mexico. Here it spreads out and turns 
eastward, passing between Florida and Cuba as the Florida 

In Florida Strait the Gulf Stream moves with a speed of 
80 nautical miles a day in the centre, and conveys about 22 
cubic miles of water an hour. The combined Florida and 
Antilles Currents move northward to Cape Hatteras with an 
average velocity of 70 miles a day in the centre, and half 
this amount on the edges and convey about 47 cubic miles of 
water an hour. Off Cape Hatteras is the so-called " Delta 
of the Gulf Stream," where it begins to break up into several 
branches. South of Nova Scotia the velocity is about 38 miles 
a day. At the south-eastern and southern edge of the Grand 
Banks of Newfoundland the Gulf Stream comes into conflict 
with the cold southward-flowing Labrador Current, which 
greatly lowers its temperature. The average temperature of 
the water flowing towards Europe after passing the Grand 
Banks is 10-15 F. lower than the temperature of the Gulf 
Stream off Cape Hatteras, and the greater part of this cooling 
must be due to the Labrador Current, either directly by 
intermingling of the warm and cold waters and the melting 
of icebergs which enter the warm current, or indirectly by 
the winds which blow from the cold to the warm water. The 
salinity of the water is also lowered somewhat. 

Off the Newfoundland Banks the Gulf Stream divides into 
several branches ; the most northerly flows towards West 
Greenland, another flows towards Iceland, and a third, 
containing the main body of the water, flows towards Europe 
and the Mediterranean. The West Greenland Current is 
felt as far as 66 N. ; it is this current which causes the decay 
of the ice brought round Cape Farewell by the East Greenland 
Current. North of 66 it appears to curve round and join the 
Labrador Current, but part of it may continue northward 
beneath the surface and cause the " North Water," the .wide 
sheet of navigable water found in the upper end of Baffin Bay 
in summer and autumn. The branch of the Gulf Stream 
which passes directly towards Iceland usually reaches the 
south-west coast, where it ameliorates the climate somewhat, 
after which it is lost. 

The third or main branch of the Gulf Stream passes directly 


eastward, again dividing in about 45* N., 40 W, into two 
branches. The southern branch turns south-eastward, skirting 
the coasts of South-west Europe and Africa as a cold current, 
and ultimately re-entering the North-east Trade Current. 
Between Gape Verde and Gibraltar, and even farther north 
in summer, it sets off the coast, and is separated from the land 
by a belt of cold upwelling water, to which we will refer again 
later. The northern branch, gaining renewed velocity from 
the prevailing south-west winds, crosses the Atlantic with an 
average speed of 12 miles a day, and bathes the shores of 
Western and North-western Europe from the Bay of Biscay 
to the North Sea. This current is banked up against the 
coast, greatly ameliorating the climate. The bulk of the water 
passes north of Ireland and Scotland to the North Sea, from 
which one arm passes west of the Faroes to Iceland, turning 
east again north of Iceland and mingling with a south- 
easterly branch of the East Greenland Current in a series of 
great whirls. 

The larger arm, under the influence of the prevailing 
southerly winds, drifts along the Norwegian and North 
European coasts to Novaya Zemlya, where it is largely 
overlain by colder but fresher water and loses its identity. 
From North Cape an arm goes northward to Spitsbergen, 
where it mingles with the westward-flowing Arctic Drift 
in another series of whirls ; this branch gives rise to the 
relatively favourable climate of Spitsbergen, and keeps the 
western coasts of this archipelago almost free of ice. In the 
Faroe-Shetland channel the volume of the warm current 
is about 2 cubic miles an hour, and it has decreased to less 
than i cubic mile off the Lofoten Islands. Even this amount, 
small compared with the volume of the Gulf Stream off the 
Atlantic coast of Florida, is of very great climatic importance, 
and its variations from year to year have important effects 
on the Norwegian harvests. An increased volume and high 
temperature (the two usually go together) of the Atlantic 
Current off Norway in May gives good harvests in the autumn 
of the same year, and diminishes the amount of drift ice in 
the Barents Sea one or two years later. 

We have seen that in the Arctic Ocean the last remnants 
of the Gulf Stream are finally lost beneath a layer of colder, 
fresher water. The latter originates chiefly in the great 


rivers of Eurasia and North America which discharge into 
the Arctic Ocean, and this surface stratum, in which ice- 
formation is very active, forms the mainspring of the return 
cold circulation. Passing north of Spitsbergen it continues 
towards the east coast of Greenland ; the main mass of the 
water follows this coast southwards as the East Greenland 
Current, bearing great quantities of ice. Owing to the 
earth's rotation, this current is banked up against the coast ; 
it rounds Cape Farewell and passes up the west coast of 
Greenland as far as Disco Island. Here it turns westward 
under the influence of the prevailing easterly winds, and 
finally, mingling with the West Greenland Current, it flows 
southward as the Labrador Current, gaining important 
accessions from Smith Sound and other channels in the 
Northern Archipelago. Off the Newfoundland Banks the 
Labrador Current meets the Gulf Stream, as we have seen, 
and helps to lower its temperature. When the Labrador 
Current and Gulf Stream meet, their densities are approxi- 
mately the same, and they mix along the junction. The 
mixture is, however, slightly heavier than either of the original 
currents, and this produces a " density wall," on either side 
of which the currents are opposed. The maximum density 
is from 20 to 30 miles inside the cold wall, so that there is a 
cold current flowing alongside the Gulf Stream in the same 
direction. There are also continual eddies breaking off from 
the cold wall and drifting eastwards. 

Before leaving the subject of the North Atlantic and Arctic 
circulation it is necessary to emphasise the part played by 
floating ice. It has been pointed out that cold water can only 
remain at the surface above warmer water by virtue of being 
lighter, because it is less saline. This relative freshness can 
be brought about in three ways : by heavy rainfall, as in the 
doldrums, by great rivers, or by the addition of ice. Ice, 
even when formed in the sea, is fresh, and though some salt 
water is usually mixed up with the ice at first, this tends to 
drain out. Hence, when the ice melts it decreases the salinity 
of the neighbouring sea water. A thin layer of relatively 
fresh water is constantly gaining salt from below by mixing 
and diffusion, and unless it continually received accessions 
of fresh water, by the time it had travelled a thousand miles 
or so it would differ little in salinity from the underlying 


water. Hence it could no longer exist as a cold surface 
current. There are not likely to be great differences of rainfall 
over adjacent parts of the ocean (if there were, the heavier 
rainfall would most likely be over the warmer water), and the 
accession of river water is only possible near the coast, so 
that the only way in which a current can remain relatively 
fresh while traversing the open ocean is by bearing with it 
large quantities of floating ice, the melting of which con- 
tinually renews the surface layer. In the absence of ice, 
the current would either lose its identity, or become heavy 
and sink below the surface. Thus, for instance, in the Arctic 
Ocean the fresh water from the great rivers is conserved 
by being frozen, instead of mixing with the more saline 
underlying water, and helps to form the floating ice-cap or 
Palaeocrystic ice. The East Greenland Current is initiated 
by this floating ice-cap and maintained by the ice which it 
carries with it, and which gradually melts. In the same way 
the Labrador Current is supplied partly by the remains of the 
East Greenland Current and partly by the ice from the 
innumerable channels of the Arctic Archipelago of America. 
It seems highly probable that if there were no floating ice in 
the Arctic Ocean the East Greenland and Labrador Currents 
would not exist ; the water as it cooled would sink, and the 
return to lower latitudes of the water brought by the warm 
currents would take place not at the surface, but below the 
surface, if not actually at the bottom. 

I have described the North Atlantic circulation, with its 
extension into the Arctic, in some detail, because a good 
grasp of it is necessary in order to understand the way in 
which changes of the oceanic currents controlled the warm 
periods. The other oceans may be dismissed more briefly. 
The western halves of the North Pacific, South Atlantic, 
and, to a less extent, the South Pacific all have warm currents 
resembling the Gulf Stream. In the North Pacific the warm 
current is unable to penetrate the Bering Strait, and there- 
fore turns south-eastward and washes the western coast of 
North America. In the South Atlantic and South Pacific 
the warm currents enter a great stream of water which 
circumnavigates the globe in the Southern Ocean, picking 
up ice from the Antarctic and sending branches northward 
along the western coasts of all the continents. In the tropical 


Indian Ocean the currents are largely controlled by the 

I have referred to the importance of upwelling cold water. 
Cold water, owing to its greater density, tends to sink towards 
the bottom, so that, except in the presence of ice, there is 
generally a steady decrease of temperature as we go deeper 
below the surface. Since the maximum density of sea water 
occurs at its freezing point of about 28 F., so long as there is 
a plentiful supply of water at this temperature the bottoms of 
the great oceanic basins will be occupied by water not much 
above the freezing point. The warm surface layer may be 
likened to a skin, and wherever this skin is broken , the colder 
underlying layers will be exposed. This will happen whenever 
there are winds over adjacent areas blowing away from each 
other (divergent winds), or whenever the winds blow off 
the coast. It will also happen whenever a current flowing 
along a coast-line is deflected away from it by the earth's 
rotation. The latter happens with the currents on the west 
coast of South America and South Africa the Humboldt 
and Benguela Currents which turn away from the coast 
and cause belts of upwelling cold water to form between 
them and the shore. The low temperature of these currents 
is due quite as much to this upwelling of cold water as to 
the original low temperature of the surface water. The 
temperature of the surface layers is also lowered slightly by 
breaking waves, which mix up the surface " skin " with the 
underlying colder layers and so cause a diffusion of heat 
through the whole depth affected by the waves. 

Along the edges of an ocean current travelling across 
the open sea there is usually a certain amount of eddy motion. 
This is especially noticeable off the Newfoundland Banks, 
where the Gulf Stream meets the Labrador Current. This 
must result in a certain amount of mixing, and a decrease 
in the volume or temperature of the warm current. It also 
lowers the average velocity, and therefore increases the* loss 
of heat by radiation and conduction while the current is 
travelling a given distance. The loss both of volume and of 
heat is greater in proportion from a weak current than from 
a strong one. Thus we may sum up the causes which lead 
to the decrease of temperature or volume in a warm ocean 
current as follows : 


1. Mixing with colder surface water by eddy motion along 
the edges. 

2. Melting of floating ice which drifts on to the warm 

3. Mixing with colder underlying water by 

(a) upwelling due to divergent winds or motion 
directed away from a coast ; 

(b) breaking waves. 

4. Cooling by conduction to the air, especially to cold winds. 

5. Cooling by radiation. 

We have next to consider the variations which these factors 
may have undergone in the geological past, and especially 
during the warm periods. In Chapter II. we found that 
during the warm periods, on the poleward sides of the sub- 
tropical high-pressure maximum, the winds tend to blow 
directly towards the poles over the whole surface of the ocean. 
These winds would act on the water in the way described in 
the first paragraph of this chapter ; that is, they would drive 
a body of water towards the right (in the Northern Hemisphere) 
or towards the eastern shores of the ocean. There would be a 
piling up of the water in the east, so that the surface of the 
ocean would slope downwards towards the west. In the 
steady state this would give rise to a wide oceanic surface 
current directed from south to north, in the same direction 
as the winds. There is no reason to suppose that the inter- 
tropical circulation during the warm periods differed from 
that found at present, and the warm currents due to the winds 
of middle latitudes would be reinforced by the warm inter- 
tropical water driven westward in the Equatorial Currents 
and rounding the western ends of the sub-tropical highs. 
Under these conditions, and with the complete or almost 
complete absence of floating ice, the occurrence of adjacent 
are^s of water at different surface temperatures would be 
reduced to a minimum. Thus, the cooling under headings 
i, 2, and 4 would be much less than at present. 

The temperature of the water at the bottom of the deep 
oceanic basins cannot be lower than that of the coldest part 
of the sea surface, in fact, owing to earth heat, it must be a 
few degrees higher. During the warm periods, therefore, 


when the surface waters of the polar oceans were well above 
freezing point, there must have been a corresponding rise in the 
temperature of the bottom layers. This implies a marked 
decrease in the vertical temperature gradient, and while 
there must always have been upwellings of underlying water, 
due to off-shore winds and currents leaning away from the 
coasts, their cooling effect must have been much less than at 
present. Further, with the decrease of storminess consequent 
on the absence of the polar fronts, steady light or moderate 
winds would prevail in middle latitudes, and there would be a 
great diminution of divergent winds and of wave motion. 
Thus the cooling of surface ocean currents under headings 
3(0) and 3 (b] would also be less than now. 

Finally, we come to 5, cooling by radiation. As will be 
seen in Chapter VI., the higher temperature of the air implies 
a greater amount of water vapour, especially above the 
oceans, but probably not an increase in the cloudiness, and 
this means that a larger part of the earth's radiation would 
be absorbed by the air, part of it being returned to the surface 
of the sea and helping to maintain the temperature. Thus 
we see that during the warm periods all the circumstances 
worked together to maintain the temperature of the warm 
ocean currents into high latitudes. Since these warm currents 
were also accompanied by warm winds, it will be seen that 
with large, open oceans and low, level continents, the extension 
of warm temperate oceanic climates into the immediate 
neighbourhood of the poles does not involve any insuperable 

An investigation of the probable systems of ocean currents 
in the northern hemisphere during the various geological 
epochs was made by P. Lasareff (4). He placed plaster 
models of continents in a circular plane basin filled with water, 
and directed streams of air obliquely towards the circum- 
ference to represent the trade winds. When the model 
reproduced the present land- and sea-distribution, the currents 
produced resembled existing currents even in detail. The 
horizontal temperature gradient was simulated by passing a 
heating coil round the edge, which represented the equator. 
The results are of great interest ; in the models representing 
the warm periods ocean currents passed across the pole, 
whereas in the cold periods no current crossed the pole. 


The results were especially effective in reproducing the 
variations of climate in Europe. The currents shown by the 
models agree with the directions of migration of marine 

The picture we have drawn of the oceanic circulation 
during the warm periods warm currents extending from 
shore to shore of the oceans and steadily drifting poleward, 
to return to low latitudes beneath the surface is not the 
only one which has been presented. T. C. Chamberlin (5) 
has arrived at very different conclusions ; he supposes that 
during the warm periods there was very great evaporation 
in low latitudes, so great in fact that the increased salinity of 
the water caused it to become heavy enough to sink to the 
bottom. Here it travelled slowly north and south towards the 
poles, retaining its heat, and rising to the surface in high 
latitudes, where it caused highly favourable climates. An 
illustration of this type of circulation on a small scale has been 
described earlier in this chapter (the " North Water " in 
Baffin Bay). But I do not think that this " reversal of the 
oceanic circulation " is a practicable explanation of climatic 
changes, for at least two reasons. In the first place, it will be 
seen in Chapter VI. that greater warmth does not necessarily 
mean greater evaporation ; once the air is saturated, it 
cannot take up any more moisture unless either the tem- 
perature rises still further or some of the water vapour it 
already contains is first condensed as rain. The temperature 
cannot go on rising indefinitely, and the conditions during 
the warm periods were less favourable to rainfall in low and 
middle latitudes than at present ; hence, evaporation was 
probably less active rather than more active than now. 
Secondly, there does not seem to be any obvious reason why 
the saline water should rise at the poles when it got there. 
In the absence of a wind-driven circulation, we should expect 
the heavy water to remain at the bottom and accumulate 
there, until it occupied all the oceans except a thin surface 
layer. Here there would be a slow drift of water from the 
regions where precipitation exceeded evaporation, and from 
the mouths of the great rivers, to the regions where evaporation 
exceeded precipitation, the drift being just enough to maintain 
equilibrium. The example of the " North Water " is not to 
the point ; the warm water here comes to the surface either 


in an eddy or because the prevailing off-shore winds drive 
away the surface layer of colder but fresher water. 

During the Quaternary Ice-Age the warm currents stood 
less chance than now of carrying an appreciable portion of 
their original warmth to high latitudes. Owing to the 
enormous quantities of floating ice which existed in the oceans, 
and which have left traces of their existence in the submarine 
accumulations of glacial material which have been dropped by 
icebergs, the surface waters must have been very cold. We 
have evidence of great icebergs in the English Channel, 
which dropped boulders weighing many tons on to the sea- 
floor at Selsey, and there are glacial accumulations off the 
west coast of Ireland and in many other localities. The 
winds from the glacial anticyclones must have driven these 
icebergs and their cold thaw water far across the oceans. 
This water, being light because of its freshness, spread over 
the warmer but more saline water of the warm currents in 
mid-Atlantic, just as it does to-day in the Arctic, so that the 
Gulf Stream, for instance, must have lost its identity in 
relatively low latitudes. 

In addition to Chamberlin's hypothesis referred to above, 
changes in the oceanic circulation induced by alterations of 
the land and sea distribution have often been suggested to 
account for climatic changes. F. Kerner has been especially 
active in explaining the warm periods in this way, but I am 
deferring a consideration of his work until Chapter VIII., 
since his method is to analyse the distribution of temperature 
resulting from the present land and sea distribution, and to 
apply the results to the geography of former geological epochs. 
We may refer here to a very old idea that the Quaternary 
Ice- Age was brought about by the omission of the Gulf Stream 
from the economy of the North Atlantic. Four ways have been 
suggested in which this may happen ; the opening of a wide 
gap between North and South America by the submergence 
of the Isthmus of Panama, allowing the Gulf Stream proper 
to pass into the Pacific instead of being bent back into the 
North Atlantic ; the northward extension of the eastern 
shore-line of South America in such a way as to deflect the 
greater part of the Equatorial Current to the south instead 
of to the north ; an increase in the velocity of the North-east 
Trades in the Atlantic relatively to the South-east Trades, 


shifting the whole system of Equatorial Currents southward 
with the same result ; and the formation of an extensive 
" Antillean Continent " across the path of the Gulf Stream 
and Antilles Current, forcing them to pass eastward much 
farther south than at present and form a closed circulation 
in tropical and sub-tropical regions. Any one of these changes 
might modify the surface circulation in the North Atlantic 
and introduce corresponding climatic changes in eastern North 
America, and especially in Europe, and we must examine them. 
The separation of North and South America by a strait 
across the Isthmus of Panama occurred during the greater 
part of the Tertiary period, and was responsible for the 
great difference in the faunas of these continents, but it does 
not appear to have persisted into the Quaternary. Hence 
from this factor alone we should have expected a cold climate 
in North-west Europe during the Tertiary, becoming warmer 
in the Quaternary, which is the reverse of what actually 
happened. Evidently the opening and closing of this gap 
did not greatly affect the Gulf Stream. South America 
stands in the course of the South Equatorial Current like a 
mighty wedge, with its apex at Cape San Roque in 5 S., and 
all that portion of the current which lies between the equator 
and 5 S., and which would normally be deflected southwards 
by the earth's rotation, is turned to the northward, and enters 
the North Atlantic as the Guiana Current. It is the Guiana 
Current which mainly supplies the warm water in the Gulf of 
Mexico. If, owing to geographical changes, the apex were 
shifted two degrees farther north, the amount of water which 
it deflects from the Southern to the Northern Hemisphere 
would be decreased by about forty per cent., an event which 
would appreciably affect the warmth of the North Atlantic. 
The north-eastern part of South America appears to have 
been slightly lower during the Quaternary than at present, 
but the configuration both of the land surface and of the 
ocean floor is such that a change of a thousand feet or more, 
whether elevation or depression, would make very little 
difference in the latitude of the apex of the wedge. There is, 
therefore, no reason to suppose that the geographical changes 
in this region during and since the Quaternary have been 
sufficiently great to introduce any important modifications 
in the volume of the Guiana Current. 


There is a considerable amount of evidence that during 
at least the early part of the Quaternary period the Gulf of 
Mexico was largely dry land ; this would merely turn the 
waters of the Guiana Current north into the Antilles Current, 
and would not greatly affect the temperature of the Gulf 
Stream off the east coast of the United States. In fact, so 
far as the geography of the Quaternary can be reconstructed, 
it was as favourable as the present for the existence of a 
powerful warm current in the North Atlantic. 

This leads us to ask if the effect of minor geographical 
changes on the great oceanic circulations has been over- 
estimated. When the project of a Panama Canal was first 
mooted, there was some popular outcry that it would allow 
the Gulf Stream to pass through into the Pacific and so 
interfere with the climate of Europe. This fear was quite 
unnecessary, but it illustrates the importance which the 
" man in the street " attaches to the slender barrier of the 
Isthmus of Panama. Actually, as we have seen, only about 
one-third of the water which forms the Gulf Stream off the 
east of Florida passes through the Gulf of Mexico at all ; 
two-thirds of it is derived from the Antilles Current, which 
takes the whole of the water from the North Equatorial 
Current. But the real reason for the existence of the Antilles 
Current is not the chain of islands known as the Antilles, it 
is the limitation of the sub-tropical antic yclonic centres to the 
eastern halves of the oceans, combined with the rotation of 
the earth, which deflects the North Equatorial Current to 
the right, i.e., northwards. 

In the North Pacific there is a gap between the Philippine 
Islands and China which is wide open to the waters of the 
North Pacific Equatorial Current, but the latter ignores 
the invitation, and instead turns northward in the open 
ocean to form the warm current which gives its favourable 
climate to Japan. In fact, under normal conditions, water 
which is travelling westward north of the equator must turn 
north, and water which is travelling westward south of the 
equator must turn south, unless hindered from doing so by 
some geographical obstacle. This flight from the equator 
will take place most readily where the isobars also trend 
away from the equator at the western ends of the sub-tropical 
oceanic anticyclones. 


Finally, there remains the fourth consideration, the unequal 
strength of the Trade winds in the two hemispheres. At 
present the atmospheric circulation over the whole Southern 
Hemisphere is stronger than that over the Northern Hemi- 
sphere, partly because the greater area of the oceans leads to 
a smaller loss of energy through friction, and partly because, 
owing to the very low temperatures over Antarctica, the 
temperature gradient between equator and pole is greater in 
the Southern Hemisphere. Hence the South-east Trades 
are the stronger, and owing to their momentum are able 
to blow right across the equator into the Northern Hemisphere. 
The difference is greatest in June to July arid least in December 
to January, but the doldrums lie north of the equator through- 
out the year. Hence the greater part of the Equatorial 
Current, both in the Atlantic and Pacific Oceans, lies north 
of the equator, and is deflected northward by the earth's 

Now we know that the glaciation of the Antarctic Continent 
began during the Tertiary, while the Northern Hemisphere 
was still enjoying genial climates in high latitudes. Hence 
we may suppose that at this period the South-east Trade 
crossed the equator to an even greater extent than at present, 
and that this helped to maintain the temperature of the 
Northern Hemisphere and to depress that of the Southern 
Hemisphere. Then the Northern Hemisphere also became 
glaciated, and, owing to the greater land area in the glaciated 
regions, these northern ice-sheets outweighed the southern, 
and caused the North-east Trades to become as strong as or 
stronger than the South-east Trades. This caused the 
Northern Hemisphere to receive a smaller share of the 
equatorial warm water, intensifying the glacial conditions in 
that hemisphere still further, while the glaciation of the 
Southern Hemisphere remained relatively slight. Hence we 
see that during the Quaternary period the variations of the 
oceanic circulation must have tended to exaggerate the 
climatic oscillations in the Northern Hemisphere and moderate 
them in the Southern Hemisphere. 

The partial closing of the gap between Greenland and 
Europe by the elevation of the submarine ridge which passes 
through Iceland and the Faroes to Scotland, which occurred 
during the Quaternary Ice-Age, must have deflected the 


Gulf Stream Drift into lower latitudes and displaced the 
Icelandic minimum southwards, altering its alignment to 
west-east or even north-west-south-east, instead of south- 
west-north-east as at present. This must have profoundly 
modified the climate of the countries bordering on the North 
Atlantic, and probably increased the severity of the glaciation 
in these regions. These changes have been discussed by the 
late F. W. Harmer (6, 7), who attempted to reconstruct the 
pressure distribution and storm tracks which would prevail 
under these conditions. His papers were written before the 
work of G. de Geer on annual clay varves had demonstrated 
the contemporaneity of at least the Wurmian glaciation in 
North America and Europe, and he supposed that the 
glaciations of these two continents alternated, but I think 
the pressure distribution which he deduces would have 
favoured increased winter snowfall over the north-eastern 
parts of North America, and hence brought about glaciation 
rather than deglaciation. 

H. J. E. Peake and H. J. Fleure (8) in some comments 
on Harmer's papers suggest a complete explanation of the 
Quaternary glacial sequence in Europe in terms of the elevation 
and depression of a Labrador-Greenland- Iceland-Scotland 
land-bridge. They point out that with this bridge complete 
there would be little or no ice in the North Atlantic in summer, 
and the climate of the British Isles would be dry and sunny, 
similar to that of the coast of British Columbia, and un- 
favourable for glaciation. A somewhat smaller elevation, 
however, which left some gaps in the land-bridge, " would 
probably increase the amount of ice in the North Atlantic 
so long as the northern lands remained much higher than at 
present," and would cause a deterioration of the climate of 
Western Europe. They point out that the first or Guiizian 
glaciation was limited to Scandinavia and the Alps, and did 
not extend to France, the North Sea, or the English plain, 
so that it might well have been the result of elevation only, 
and they suggest that during this glaciation the land-bridge 
was complete and the climate of England favourable. The 
cold period in Eastern England, represented by the Weybourne 
Crag and Chillesford Beds, which appears to correspond with 
the Gunzian glaciation, would then fall either just before or 
more probably just after the time of maximum elevation. 


The later glaciations occurred during periods of lesser elevation, 
when the land-bridge was not complete and there was much 
ice in the North Atlantic ; the interglacial periods occurred 
during the intervals of subsidence in which the land fell below 
its present level. 

There is quite a lot to be said for this interpretation of 
the Quaternary sequence in Europe. The suggestion that 
a large amount of cold water was accumulated in a closed 
Arctic basin, and that the level of the Arctic Ocean may 
even have risen well above the general level of the remaining 
oceans, this mass of cold water being subsequently released 
by a depression of the land and flooding southwards into 
the Atlantic, may be especially fruitful. It gives a plausible 
explanation of the sudden appearance of the Arctic fauna 
in middle latitudes, for example, in the Sicilian (3OO-foot) 
raised beaches of the Mediterranean, which contain a fauna 
now found only in the northernmost parts of Europe. But 
we must not forget that the Quaternary sequence in Europe 
was paralleled by that in other parts of the world, such as the 
Himalayas, and that any explanation of the phenomena 
found in Europe must fit into place in a larger scheme which 
takes account of the whole world. 

The final role of the oceans in climatic changes to which 
we have to refer is that of regulator. It is well known that 
owing to the high specific heat of water and to the fact that 
the changes of temperature penetrate to a greater depth, 
a large sea or ocean takes much longer to warm or to cool 
than does a land surface in the same latitude. Hence the 
annual range of temperature on islands or windward coasts 
is much less than that in the interior of great continents. But 
this conservation of heat is not limited to periods of a year 
or even a few years. During a period of cooling climate, 
the coojed sea water sinks to the bottom of the oceans and the 
warmest water remains at the top. 

If the ocean covered the whole surface of the earth it 
would have an average depth of 2,600 metres (8,500 feet) ; 
if we take the amount of heat reaching the outer limit of 
the earth's atmosphere from the sun as 720 calories per 
square centimetre per day (see next chapter), we find that 
the whole of this solar heat for a year would have to be 
absorbed and retained by this universal ocean, to raise the 


temperature by one centigrade degree. Again, suppose 
that at the end of a long warm period the mean temperature 
of the oceans is 10 C. (18 F.) higher than at present. When 
we remember that a large portion of the oceanic water is 
now little above freezing point, it is seen that this is not an 
exaggerated assumption. Then if for some reason the 
equilibrium temperature sank to its present level, the heat 
conserved in the oceans would suffice to maintain the average 
temperature 2 C. (3-6 F.) above the present for a period 
of nearly 250 years. The retardation of warming up after 
a cold period would be less effective, since the cold water 
would remain at the bottom of the ocean without greatly 
affecting the higher layers. 

R. Spitaler (9) goes even further than this. In his theory 
of the astronomical cause of ice-ages (Chapter V.), he attempts 
to get over the difficulty that increased eccentricity of the 
earth's orbit would act oppositely in the Northern and Southern 
Hemispheres by supposing that the regulating effect of the 
oceans could maintain glacial conditions during a period of 
10,000 years while conditions were otherwise not specially 
favourable for glaciation. He does not give any calculations 
in support of this figure, and it seems to be excessive. Spitaler's 
claim would perhaps have been based more soundly on the 
consideration that owing to the creep of cold water along 
the ocean floor, severe glaciation in one hemisphere would 
suffice to maintain low temperatures throughout the bottom 
layers of the whole ocean, and so to some extent lower the 
temperature in the other hemisphere also, though the process 
would not be very effective. 

Huntington and Visher (10) suggest that the growing 
salinity of the oceans during the course of geological time 
may have had some climatic effect. This may be so, but I 
doubt if the effect can have been noticeable during the greater 
part of geological time. The accession of salt to the ocean is 
at present derived almost entirely from the sedimentary rocks, 
that is, it has previously been withdrawn from the oceans. 
The very great estimates of the duration of the pre-Cambrian 
period now current nearly a thousand million years 
suggest that even at the beginning of the Palaeozoic the ocean 
had a long history behind it, and was almost as salt as it is now. 
This is borne out by the relatively advanced organisation of 


the earliest fossils, which also suggest life in salt water rather 
than in fresh. It is possible, however, that a smaller salt 
content may have been a contributory cause in the pre- 
Gambrian glaciations. The fact that fresh water reaches its 
greatest density at a temperature above its freezing point, 
while ordinary sea water freezes before it cools to its maximum 
density, is the reason why fresh-water lakes freeze more 
readily than ocean inlets of the same depth. In very early 
geological ages it is possible that the sea was less salt than 
now, and if the difference was sufficient to bring the tempera- 
ture of maximum density above the freezing point, which 
would happen if the salinity were less than 24-7 parts per 
thousand compared with the present value of about 35 per 
thousand, the surface of the ocean would have frozen more 
readily than at present. Other conditions being equal, 
therefore, glaciation of coastal mountains would have been 
easier in very early geological ages than at present. 

The fluctuations of salinity from one geological epoch to 
another may also have affected the capacity of the air to 
absorb moisture from the oceans. The withdrawal of a great 
volume of fresh water during the glacial periods, to be locked 
up in the form of ice, must have increased the average salinity 
slightly. Moreover, there are variations in the amount of 
soluble matter locked up in salt and gypsum beds, etc. ; 
at the close of a long warm period with shallow seas and 
numerous lagoons this amount must have been appreciably 
greater than at present. Thus we may suppose that there 
have been small fluctuations of salinity, with minima at the 
end of the long warm periods and maxima during the ice-ages, 
superposed on a very slow secular increase. Decreased 
salinity would increase the vapour pressure over the oceans, 
and it will be seen in Chapter VI. that an increase of water 
vapour in the atmosphere tends to raise the mean temperature. 
In this way there would be a tendency for both the warm 
periods and the ice-ages to be intensified with also a slight 
secular fall of temperature. I think, however, that at least 
since the middle of the Palaeozoic period the variations of 
temperature due to this cause alone must have been so small 
as to be negligible compared with the other causes of variation. 
It seems highly improbable that at any stage in the known 
geological record, with the possible exception of the early 


pre-Cambrian, was the main mass of the sea water sufficiently 
fresh for its temperature of maximum density to be above 
its freezing point. 


(1) DURST, C. S. " The relationship between current and wind." London, 

Q,. J. R. Meteor. Soc., 50, 1924, p. 113. 

(2) HEPWORTH, M. W. CAMPBELL. " The Gulf Stream." London, Geogr. J., 

1914, p. 431. 

(3) SVERDRUP, H. U. " Oceanography for meteorologists." New York, 1942. 

(4) LASAREFF, P. " Sur un methode permettant de demon trer la d^pendance 

des courants occaniques des vents aliz^s et sur le role des courants 
oceaniques dans le changement du climat aux epoques gcologiques." 
Beitr. Geoph., 21, 1929, p. 215. 

(5) CHAMBERMN, T. C. " An attempt to frame a working hypothesis of the 

cause of glacial periods on an atmospheric basis." J. GeoL, Chicago, 
7> '899, PP. 545 and 667. 

(6) HARMER, F. W. " The influence of the winds upon climate during the 

Pleistocene epoch." London, Q.J. Geol, Soc., 47, 1901, p. 405. 

(7) HARMER, the late F. W. " Further remarks on the influence of the winds 

upon climate during the Pleistocene epoch." London, Q,. jf. R. Meteor. 
Soc., 51, 1925, p. 247. 

(8) PEAKE, HAROLD J. E., and H. }. FLEURE. " The Ice-Age." Man, 1926, 

p. [4]- 

(9) SPITALER, R. "Das Klima des Eiszei takers." Prag, 1921. (Litho- 


(10) HUNTINGTON, E., and S. S. VISHER. " Climatic changes, their nature 
and cause." New Haven, 1922. 



IT has been shown in the preceding chapters that a com- 
paratively small initial change in the mean temperature 
of the polar regions might be so magnified by secondary 
effects, especially in connexion with the polar ice-caps, that 
the final result would be a very great change of climate, 
sufficient to account for the genial polar climates of the 
" warm " periods. We have now to begin our search for such 
possible initial causes, and the most obvious place to look is 
the great source of all warmth and life on the earth the sun. 
Careful studies are being made, especially by the Astrophysical 
Observatory of the Smithsonian Institution (i), of the amount 
of heat radiated by the sun, and of the variations to which it is 
subject. The measurement aimed at in the first place is the 
amount of heat which would reach a unit area of the earth's 
surface exposed to the solar beam at right angles, if none 
of it were intercepted by the earth's atmosphere. As it is 
impossible to get outside the atmosphere to obtain these 
measurements, the result has to be arrived at indirectly. 
Observatories have been established at several points on high 
mountains, where the air is normally very dry and the sky 
clear ; the chief of these observatories are at Montezuma in 
Chile (8,895 feet), Table Mountain, California (7,500 feet) 
and Mount St. Katherine, Egypt (8,500 feet). Elevated 
stations are chosen because the mass of air through which the 
sun's rays have to penetrate is less ; dry regions because part 
of the solar radiation is absorbed by water vapour and also 
because a clear sky is essential for regular observations. At 
these places the direct heating power of the sun is therefore 
greater than it is on low, humid and cloudy plains, but there 
is still a large amount of absorption. This is calculated in 
two ways. When the sun is nearly vertical, the thickness of 
air through which its rays have to penetrate is much less than 
when it is near the horizon, and the heat received is 


consequently greater. Making a number of observations at 
intervals during a morning when the meteorological conditions 
remain practically uniform is therefore almost equivalent to 
making observations at the same moment at different heights 
above the ground. It is found that with radiations of certain 
wave-lengths which are only absorbed slowly by the 
atmosphere, the logarithm of the amount lost is proportional 
to the air mass through which the rays have passed, and it is 
possible to calculate the amount of heat which these rays would 
deliver if there were no atmosphere. By means of a spectro- 
scope these measurements are taken for a number of different 
wave-lengths, and the results are plotted. In general, they 
show a smooth curve which is, however, interrupted by a few 
marked depressions corresponding with the wave-lengths of 
radiations which fail to penetrate the air at all. These rays 
are almost or quite absorbed in the high levels of the 
atmosphere, and in order to include them in the estimate of the 
sun's total radiation an allowance has to be calculated from 
the values of neighbouring wave-lengths. As a result of a series 
of measurements and calculations carried out by the Astro- 
physical Observatory at various stations from 1920 to 1939, the 
mean value of the intensity of the sun's radiation outside the 
limits of the earth's atmosphere is given as i 945 calories per 
square centimetre per minute. This means that a layer of 
cold water a centimetre deep exposed to a vertical sun, 
absorbing all the solar radiation and giving out none in 
exchange, would warm up at the rate of nearly 2 C. a 
minute. This value is termed the " solar constant," but the 
name is somewhat misleading, as it seems probable that the 
mean value of i 945 is subject to small variations on either 
side of the mean. 

The range of these variations is still subject to doubt, owing 
to uncertainty in the daily values. The annual means for 
the 20 years from 1920 to 1939 range from 1-928 in 1922 to 
i -950 in 1921, or just over one per cent, of the mean, but both 
of these are early values and somewhat uncertain. Abbot 
finds a number of short periodicities in the data, the longest 
being 23 years ; he estimates the amplitude of this cycle 
as less than 0-5 per cent, of the average value of the solar 
constant. It is clear that so small a change, even if maintained, 
would not suffice to bring about great changes of climate. 


The possibility of much greater changes of solar radiation in 
geological time cannot, however, be ruled out. Sir George 
Simpson (2, 3) has discussed the probable effects of a large 
oscillation of solar radiation, and has shown that they agree 
with the observed climatic changes during the Quaternary. 

The first effect of an increase in the solar radiation would 
be to raise the temperature everywhere on the surface of the 
earth, but more in low than in high latitudes. This would 
immediately increase the amount of evaporation from water 
surfaces, and also the strength of the atmospheric circulation. 
More evaporation means more cloud and more precipitation. 
But as will be seen in Chapter VII. , cloud reflects back to 
space a large part of the solar radiation which falls on it. 
Hence an increase of cloudiness lowers the temperature. 
The final result of a large increase of solar radiation would 
therefore be a slight rise of temperature and a great increase 
in cloudiness and precipitation. The increase in total 
precipitation with increasing radiation is shown in curve I of 
Fig. 9, reproduced by permission of Sir George Simpson 
and the Manchester Literary and Philosophical Society 
from (2). 

In high latitudes and especially on high ground, a large 
part of the total precipitation falls as snow. As the radiation 
increased, the proportion of precipitation falling as snow 
would decrease, but for a time this decrease would be slower 
than the increase of total precipitation. Hence the snowfall 
would increase at first, but with a further increase of radiation 
the general rise of temperature would cause so much less of 
the precipitation to fall as snow that the total snowfall would 
begin to decrease. The curve of snowfall is shown as II in 
Fig. 9- 

The accumulation of snow to form ice-sheets results from 
the excess of snowfall over melting. The melting is shown in 
curve III ; so long as the summer temperature does not rise 
above freezing point it is inappreciable (point A on the curve) . 
With rising temperature it grows slowly at first, so long as the 
snow cover is complete, but as soon as the melting is sufficient 
to expose part of the underlying surface it proceeds rapidly. 
At point B the curves of snowfall and melting intersect and 
above that point any accumulation of snow or ice will dis- 
appear. Curve V shows the accumulation of snow during 


a period of increasing radiation. With decreasing radiation 
the changes would proceed in the reverse order. 

In low latitudes where there is no snow on low ground the 
result of increasing radiation would be a progressive increase 
of rainfall, but on equatorial mountains of sufficient height 
there would be a glacial cycle resembling that in high latitudes. 


Fig. 9. Effect of increasing radiation on precipitation and 
accumulation of snow. 

Fig. 10, also reproduced by permission from (2), shows the 
effect of a double oscillation of solar radiation. The first 
period of increasing radiation causes an ice-age in high 
latitudes and increasing rainfall in low latitudes. At the 
maximum radiation the ice melts and we have a warm moist 
interglacial period in high latitudes coinciding with the peak 
of a pluvial period in low latitudes. As radiation decreases, 
there is a return of glaciation at first, but as radiation and 



precipitation decrease still further, the latter, even though it 
falls entirely as snow, is insufficient to balance the loss of ice 
by outflow and the ice-sheets finally disappear. We now 
enter a long cool arid interglacial period, which lasts until 
a new increase of solar radiation brings about a renewal of the 
glacial and pluvial cycle. 




Fig. 10. Effect of two cycles of solar radiation on glaciation. 

As shown in the lowest curve of Fig. 10, this reconstruction 
fits in admirably with the classical sequence of glacial and 
interglacial periods worked out by Penck and Bruckner (4). 
The first crest of solar radiation represents the Gunz and Mindel 
glaciations and the relatively short Gunz- Mindel inter- 
glacial. The trough of low solar radiation represents the long 


Mindel-Riss interglacial. Finally the second crest of radiation 
represents the Riss and Wurm glaciations and the short 
Riss-Wurm interglacial. 

Sir George Simpson worked out the geological and climatic 
implications of his theory iri greater detail in another paper 
(3), and showed that they are consistent with the general 
climatic and biotic history of Europe during the Quaternary. 
There are, however, certain difficulties. 

In the first place, the assumption is made that the sun is a 
variable star. If such variations are periodic, glaciation 
should have recurred at short intervals throughout geological 
time ; the theory cannot account for the long genial periods. 
Sir George Simpson tends to meet this difficulty by continental 
drift (see Chapter XIII.), but even so, such marked changes of 
precipitation as the theory requires should have left their 
records in the character of the sedimentary rocks, but there 
is no trace of such a regular cycle in the geological record. 
This difficulty was to some extent met by F. Hoyle and R. A. 
Lyttleton (5), who supposed that at intervals of time of the 
order of 100 million years the sun passes through clouds of 
interstellar matter. The particles on and near the track fall 
into the sun, their kinetic energy being converted into heat 
and giving rise to increased solar radiation. Since the cloud 
would in general be densest near the centre, radiation would 
rise to a maximum and then decrease again, the time of the 
passage being of the order of 100,000 years. Further, since 
many such clouds are irregular, the one causing the Quaternary 
ice-age may have had two centres, so giving two maxima 
of radiation and four glaciations. 

Secondly, the glacial sequence was far more complex than is 
represented in the simple fourfold scheme of Penck and 
Bruckner. The Riss glaciation had two or three distinct 
maxima and the Wurm three, in addition to retreat stadia. 
This objection is not very serious, however, because there is 
no difficulty in postulating minor but still large fluctuations 
of solar radiation superposed on the smooth curve of Fig. 10. 

The third difficulty concerns the climates of the interglacial 
periods. By the solar radiation theory the Gunz-Mindel and 
Riss-Wurm should have been mild and very wet, the Mindel- 
Riss cool and very dry. The available evidence suggests, 
however, that the general succession of vegetation was very 


similar in both Mindel-Riss and Riss-Wurm interglacials, 
indicating that the recession of the ice in Europe was followed 
in each case by a rise of temperature to a level somewhat 
higher than the present. This rise in turn gave place to a fall 
and finally to the onset of another glaciation. The post- 
glacial period has so far followed a similar course, a rise to a 
maximum in the " Climatic Optimum " followed by a fall, 
but it is too early yet to say that we are now in the second 
half of an interglacial period. 

Finally, the available data do not support the hypothesis 
that in low latitudes one pluvial period represents two 
glaciations. The pluvial sequence has been worked out in 
detail in tropical Africa, where the rises and falls of the great 
lakes are believed by E. Nilsson (6) to have been contem- 
poraneous with the advances and retreats of the mountain 
glaciers. For example, in the last interpluvial period, when 
the prehistoric Lake Kamasia dried out completely, the 
mountain glaciers also melted completely. Subsequent oscilla- 
tions of climate also run closely parallel in both lakes and 

The evaporation from the tropical oceans, and hence the 
total rainfall over the globe, depend on the wind velocity as 
well as on the temperature. Large ice-sheets in high latitudes 
increase the temperature gradient and hence the strength of 
the atmospheric circulation, while at the same time, owing 
to the existence of large semi-permanent glacial anticyclones 
and the equatorward deflection of the storm tracks, they 
decrease the area over which most of the rain falls. Hence, 
other things being equal, each glacial period must have been 
a pluvial period in tropical and sub-tropical regions and each 
interglacial an interpluvial period. Whether this oscillation 
was superposed on a longer oscillation in which one crest 
covered two glaciations is not yet clear. 

We may sum up by saying that if the radiation of the sun 
has varied greatly in the past the variations would have had the 
effects postulated by Sir George Simpson, but that there is 
no direct evidence of such variation, and the indirect evidence 
only partially supports the theory. 

F. Kerner-Marilaun in his book on Palaeoclimatology (7) 
takes the standpoint that variations of solar radiation should 
only be brought in as a last resource, i.e., if it is impossible to 


account for the climate of a period of geological time by the 
known factors of land and sea distribution, volcanic dust, 
etc. As an example he considers the temperature of Messel, 
near Darmstadt, Germany, during the Eocene. From geo- 
graphical and astronomical factors he calculates the probable 
temperatures as : January, 50 to 56 F., July, 70 to 74 F. 
From the plant remains and the nature of the deposit the 
temperature is estimated as 56 in January and 68 in July. 
The differences are within the limits of probable error and it 
is not necessary to bring in unknown factors such as a change 
of solar radiation. In other examples the agreement is less 
good but it is not yet possible to say whether the discrepancies 
are due to errors in the calculations, in the interpretation of 
the gcologiccil data or to changes of solar radiation. Never- 
theless the author foreshadows the time when, from a large 
number of such calculations in different geological periods, 
it may be possible to reconstruct the variations of solar 
radiation. Till then the question must be left open. 

As another example of such a calculation I may mention 
a reconstruction of the probable climatic conditions in the 
middle Eocene of south-east England which I made for 
Mrs. E. M. Reid and Miss Chandler (8). I estimated the 
probable upper limit of the mean annual temperature as 
65 F. (a remarkable agreement with Kerner-Marilaun's quite 
independent figure) whereas Mrs. Reid and Miss Chandler 
inferred from the vegetation that the mean temperature was 
probably not below 70 F., and I suggested that the dis- 
crepancy might possibly be due to increased solar radiation. 
In all other respects the climatic reconstruction agreed closely 
with the inferences from the vegetation. 

Brief reference may be made here to a suggestion by K. 
Himpel (9) that the sun suffered Nova-outbreaks in Mid- 
Algonkian (pre-Cambrian), late Upper Carboniferous and late- 
Pliocene, the average interval between outbreaks being 250 to 
350 million years, which he claims is about right for Novae. Each 
outbreak caused a thousand-fold increase in radiation which 
resulted in a corresponding increase of precipitation, followed 
by rapid cooling to a level below the present. At one stage 
during the cooling there was a maximum snowfall and con- 
sequently glaciation. Further, each outbreak diminished the 
mass of the sun and consequently its normal radiation, leading 


to a general level of climate somewhat cooler than before. 
It seems impossible, however, that even a short-lived thousand- 
fold increase of solar radiation could have occurred without 
devastating animal and plant life, and there is certainly no 
trace of such a catastrophe in late Pliocene. 

Total radiation is not the only solar characteristic which 
probably affects terrestrial climate ; we have also to take 
account of the nature of the radiation. An index to the latter 
is found in the spotted area of the sun. The spottedness is 
generally expressed as the " relative number" (10), which 
is obtained by an arbitrary formula, but which has been 
found by photographic comparisons to be closely proportional 
to the spotted area. A relative number of 100 corresponds 
with about one five-hundredth of the sun's visible disc covered 
by spots, including both umbra and penumbra. Since 1749 
the monthly mean relative sunspot numbers have varied 
between o and 206. 

The sunspot " relative number " does not bear any simple 
relation with the solar constant. Sunspots show a very marked 
and persistent oscillation of approximately 11-2 years, falling 
to an annual mean of 10 or less at sunspot minimum and rising 
to a maximum which has varied since 1749 from annual 
means of 46 to 154. This i i-year oscillation is not recognisable 
in the values of solar radiation, so it must represent a variation 
of some other solar characteristic. It seems to be an almost 
permanent characteristic of the sun, for a cycle of ten or 
eleven years has been found in the thickness of annual layers 
in laminated deposits of various geological ages, including the 
Upper Carboniferous glacial clays of Australia. 

This eleven-year oscillation is not the only one shown by 
sunspots ; more important for our purpose is the fact that 
both the eleven-year means and the heights of the maxima 
undergo large variations over decades. Thus the annual 
maxima varied from 154 in 1778 to only 46 in 1817, and back 
again to 139 in 1870. Previous to 1749 we have no systematic 
observations, but there are a number of scattered references 
in the Chinese archives dating back nearly to the beginning 
of the Christian era and a few records from Europe going back 
to medieval times. From these there appears to have been 
an important maximum of solar activity towards the close of 
the eleventh century, and another, which R. Wolf believed to 



be the absolute maximum over the whole of the Christian era, 
about 1372. 

There is little doubt that some relation exists between the 
sunspot cycle and terrestrial^conditions, but it is very obscure. 
In tropical regions temperature averages about 2 F. higher 
at spot minimum than at spot maximum ; this relation 
disappears in temperate regions but reappears in the Arctic, 
where it is clearly shown both in the temperature of Spitsbergen 
and in the amount of ice in the Barents Sea, indicating that at 
sunspot maximum the temperature falls in the Arctic and the 
area of the floating ice-cap increases. Over the world as a 
whole rainfall seems to have a tendency to be greatest at spot 
maximum, but there are many exceptions. One of the most 
interesting relations is between sunspot relative numbers and 
the frequency of thunderstorms, which are most frequent at 
sunspot maxima. The agreement is close in inland and 
tropical regions, especially Siberia, the West Indies and the 
tropical Pacific, but almost or quite vanishes in the stormy 
regions of western Europe (u). For the world as a whole 
the variation from sunspot minimum to sunspot maximum 
amounts to more than 20 per cent, of the mean number of 

These various relationships suggest that at periods of great 
solar activity as marked by large sunspot numbers, the earth 
as a whole should be somewhat cooler, rainier and more 
stormy than at times of little solar activity. Ellsworth 
Huntington and S. S. Visher (12) have based a complete 
theory of climatic change on these relationships. According 
to their view, the climate of the earth is largely governed by 
the relations between sunspots and storminess. Increased 
solar activity is considered to result in increased storminess, 
together with some displacement of the storm tracks. The 
greater vertical movement of the air associated with the 
increased storminess carries greater quantities of heat from the 
earth's surface to the higher levels of the atmosphere, where 
nearly half of it is lost by radiation to space, and when such 
a period of increased activity occurs with extensive and high 
continents, and perhaps with other favourable conditions, such 
as a paucity of carbon dioxide, a glaciation results. This 
combination of circumstances is considered to account for 
the Quaternary glaciation, and perhaps also for that of the 


Per mo-Carboniferous period. In the latter, the storm tracks are 
placed very far south, and higher latitudes are supposed to have 
remained unglaciated because they were occupied by deserts, a 
conclusion which ignores the extensive development of coal 
measures. Periods of slight solar activity and few sunspots 
had slight storminess and steady winds from low to high 
latitudes, hence they were periods of mild and equable climate 
over the whole earth. The variations of solar activity in the 
geological past are connected by Huntington with changes 
in the distance of the nearest fixed stars, especially double 
stars, a conclusion which few, if any, astronomers endorse. 

It will be seen in Chapter XXII. that the variations of 
rainfall in the temperate zone during the Christian era have 
run fairly parallel with the variations of solar activity shown 
by the records of sunspots and aurorae. But the phenomena 
of the Quaternary Ice-Age were on a scale many times greater, 
and would require enormous and prolonged outbursts of 
sunspots, which seem quite improbable. Moreover it has 
been shown (14) that the terrestrial changes are not pro- 
portional to the sunspot relative number ; the effect falls off 
rapidly as the relative number increases. Hence, while 
variations of sunspot activity may account for some of the minor 
glacial oscillations, it is unlikely that they played any appre- 
ciable part in the four main glacial advances and retreats of the 
Quaternary Ice-Age, and still less likely that they caused that 
Ice-Age as a whole. The hypothesis also breaks down 
completely over the Permo-Carboniferous glaciation. 

One other speculative work may be briefly referred to, 
because it is typical of a number of early hypotheses which 
rely on a diminution of the sun's radiation to explain the 
occurrence of ice-ages. E. Dubois (13) assumed that the sun 
has passed through a series of stages represented now by 
various fixed stars. At first, when the sun was in the stage 
represented by blue-white stars, radiation was more intense 
and the terrestrial climate was warm. The sun then changed 
to yellow, its radiation diminished, and the earth cooled. 
This was during the Tertiary period, and at intervals the 
sun passed into a third stage, intermediate between the yellow 
and red stars, during which its radiation was still feebler. 
These intervals formed the various glacial epochs of the 
Quaternary. The only " evidence " adduced in support of 


the theory concerns the frequency of blindness to certain 
colours, which is regarded as a reversion to previous geological 
epochs, the spectroscopic analysis of the light of phosphorescent 
animals, and similar peculiarities. This theory is quite 
untenable. In the first place the postulated changes in 
radiation would not have had the effects attributed to them. 
Secondly, the age of the sun is now estimated as between a 
hundred thousand million and a million million years, while 
the age of the oldest known rocks is not much more than a 
thousand million years (Appendix I.), so that it is highly 
improbable that the later stages of the geological record 
should have seen any appreciable change in the sun's constitu- 
tion of the type supposed by Dubois. Such a conclusion 
is in fact borne out by the record of the rocks. The warm 
climates of the early Palaeozoic period might have lent colour 
to a belief in a hotter sun at that time (some 400 to 500 million 
years ago), but it happens that the earliest reliable climatic 
record we possess shows us an ice-age. 

Variations in the distance of the nearest " fixed " stars have 
been suggested as a possible source of climatic changes, through 
variations in the radiation received from this source, but this 
suggestion must be ruled out for two reasons. At present 
the heat received from all the fixed stars together is not one 
ten-millionth of that received from the sun, and there is thus 
no possibility of explaining colder periods than the present 
on these lines. The close approach of a large star might 
account for a warmer period, but such an event would have 
left its traces in a derangement of the solar system, and 
astronomers say that it could not have happened at any time 
in the geological past. 

We may end this chapter with a reference to a curious 
hypothesis put forward by R. L. Ives (15) to explain the 
Permo-Carboniferous glaciation. He suggests that a small 
satellite of the earth, either original or captured, was dis- 
rupted by tidal and other forces in the late Palaeozoic to form 
a ring of fragments round the equator, similar to Saturn's 
rings. This caused low equatorial temperatures, with storms 
and heavy snowfall along the boundaries of the shadow. 
Multiple glaciation is accounted for by the occurrence of a 
series of stages in the break-up. Such an occurrence seems 
intrinsically improbable, however, and in any case it can be 


readily calculated that a ring of this nature, if it cast an 
effective shadow, would lower the temperature of the whole 
earth to such an extent as to make life impossible. 


(1) WASHINGTON, SMITHSONIAN INSTITUTION. ''Annals of the Astrophysical 

Observatory," vol. vi, 1942. By C. G. Abbot, F. E. Fowle, and W. H. 

(2) SIMPSON, G. C. " Past climates." Mem. Manchr. Lit. Phil. Soc., 74, 

i9 2 9-3> no - x > PP- 34- 

(3) SIMPSON, G. C. " The climate during the Pleistocene period." Proc. 

Roy. Soc., Edinburgh, 50, 1929-30, p. 262. 

(4) PENCK, A. and E. BRUCKNER. " Die Alpen in Eiszeitalter." Leipzig 

3 Vols., 1901-9. 

(5) HOYLE, F. and R. A. LYTTLETON. " The effect of interstellar matter on 

climatic variations." Proc. Camb. Phil. Soc., Cambridge, 35, 1919, p. 405. 

(6) NILSSON, E. " Quaternary glaciations and pluvial lakes in British East 

Africa." Geogr. Ann., Stockholm, 13, 1931, p. 249. 

(7) KERNER-MARILAUN, F. " Palaoklimatologie." Berlin (Gebr. Borntraeger) , 


(8) REID, E. M. and M. E. J. CHANDLER. " The London Clay flora.", 

London, 1933. 

(9) HIMPEL, K. " Die Klimate der geologischen Vorzeit." Veroff. Astron. 

Ges. Urania, Wiesbaden, nr. 4, 1937. 

(10) ABBOT, C. G. " The sun." London and New York, 1912. 

(11) BROOKS, C. E. P. "The variation of the annual frequency of thunder- 

storms in relation to sunspots." London, Q.. J. R. Meteor. Soc., 60, 1934, 

P- 153- 

(12) HUNTINGTON, E., and S. S. VISHER. " Climatic changes, their nature and 

cause." New Haven, 1922. 

(13) DUBOIS, E. " The climates of the geological past." London, 1895. 

(14) BROOKS, C. E. P. "Non-linear relations with sunspots." London, 

Q.. J. R. Meteor. Soc., 53, 1927, p. 68. 

(15) IVES, R. L. " An astronomical hypothesis to explain Permian glaciation." 

Philadelphia, J. Frankl. Inst., 230, 1940, p. 45. 


THE radiation emitted from the sun is not the only 
factor in determining the solar climate of the earth. 
Whether or not the total heat received by the earth 
in the course of a year has remained constant, its distribution 
among the belts of latitude during the different months has 
certainly varied from time to time, and this distribution can 
be calculated. There are three variables to be considered 
in this respect. The first is the obliquity of the ecliptic, or 
the angle which the plane of the equator makes with the 
plane of the earth's orbit round the sun. It is this which 
causes the seasons ; the greater the obliquity of the ecliptic 
the greater is the contrast between the heat received in summer 
and that received in winter. In 1910 the obliquity was 
23 27' 3-58", and it was decreasing at the rate of 0-47" a 
year, but the limits of its variation are difficult to calculate. 
Lagrange found a maximum of 27 48' in 29,958 B.C., a 
minimum of 20 44' in 14,917 B.C., and a maximum of 23 53' 
in 2167 B.C. J. N. Stockwell gives much narrower limits, 
ranging from 24 36' to 21 59', with a maximum of 24 17' in 
8150 B.C., since when there has been a steady decrease. 
Drayson, on the basis of a theory not accepted by the majority 
of astronomers, supposed the obliquity to range from 35 to 
11, the period being 31,680 years. The latest work by M. 
Milankovitch (i) assumes a variation between 22 and 24^ in 
a period of 40,400 years. 

The second variable is the eccentricity of the earth's orbit. 
This orbit is elliptical, with the sun at one of the foci, and the 
distance between the centre of the ellipse and this focus, 
expressed in terms of the major axis of the ellipse, is termed 
the eccentricity. It varies in a period of about 100,000 years 
from zero to a value of about 0-07. When the earth is nearest 
to the sun it is in perihelion, when most distant, in aphelion. 
At perihelion the earth travels along its orbit more rapidly 
than at aphelion. Thus the season which coincides with 


perihelion will be short and relatively warm, that which 
coincides with aphelion will be long and relatively cold, but 
the total amount of heat received on each hemisphere in the 
course of a year will be the same. At present, the Northern 
Hemisphere has its winter in perihelion and its summer in 
aphelion ; with the Southern Hemisphere, of course, the 
reverse is the case. Hence the solar climate of the Northern 
Hemisphere is less extreme than that of the Southern Hemi- 
sphere ; the fact that, actually, the climate is much more 
extreme in the Northern Hemisphere is due to the preponder- 
ance of land there. The season in which perihelion falls is 
not constant but undergoes a cyclic variation with a period 
of 21,000 years. Thus 10,500 years ago the Northern Hemi- 
sphere had its winter in aphelion and its summer in perihelion, 
consequently a more extreme solar climate. This regular 
variation is termed the precession of the equinoxes ; it is the third 
of our astronomical variables. 

The variation of the eccentricity and the precession of the 
equinoxes form the basis of CrolPs famous astronomical theory 
of the Quaternary Ice- Age (2). He supposed that at periods 
of great eccentricity the hemisphere with its winter in aphelion 
had a. climate so severe that, if geographical conditions were 
favourable, the snowfall during the long cold winter was 
heavy enough to persist through the short hot summer arid 
thus develop ice-sheets. At the same time the opposite 
hemisphere was enjoying a genial or interglacial period. Croll 
justly points out that the power of a snow surface in reflecting 
the sun's rays back to space without their having any warming 
effect on the earth is of great importance in the heat economy 
of ice-ages, but, apart from this, his discussion of the 
meteorological changes associated with periods of maximum 
eccentricity is probably unsound. Moreover, recent geological 
investigations have shown that the glacial periods and other 
climatic changes are practically synchronous in the two 
hemispheres, while de Geer's absolute dating shows that the 
periods at which glaciations occurred do not fit in with those 
required by droll's astronomical theory. 

Rudolf Spitaler (3) re-investigated the astronomical theory 
of climatic changes. His first step was to relate the mean 
temperature of any latitude in any month to the amount of 
heat received from the sun in that latitude. For this purpose 


he analysed the existing mean temperatures of the different 
latitudes between 60 N. and 60 S. in January, July, and 
the year, and obtained an expression for the mean tem- 
perature in any latitude in any month, which may be rewritten 
as follows, to give Fahrenheit degrees : 

'=(-17+156 S +2 9 SJ (W)-H- 3 6+26 4 S m ) (L). 

Here S is the average daily heat received on a horizontal 
surface at the limit of the earth's atmosphere in the latitude 
in question during the year, S m is the average daily heat 
received under the same conditions during the month m 
(the units being so chosen that the value of S at the equator 
during an equinoctial day is approximately 0*5), and L is 
the fraction of land and W the fraction of water covered by 
the line of latitude. By making the appropriate changes in 
the values of S and S m the temperatures under various 
astronomical conditions can be calculated from this formula, 
and by varying L and W the effect of varying land and sea 
distribution can be introduced. 

Spitaler rejects CrolPs theory that the conjunction of a long 
cold winter and a short hot summer provides the most 
favourable conditions for glaciation, and adopts the opposite 
view, first put forward by Murphy (4) and now generally 
adopted, that a long cool summer and short mild winter are 
the most favourable. Spitaler's attempt was rational, but 
fails to fit in with the actual sequence of events ; in particular 
his time-scale is hopelessly impossible. For example he ends 
the Wurm glaciation at 89,680 B.C. whereas geological 
evidence puts it at a mere 18-20,000 B.C. 

W. Koppen and A. Wegener (5), while accounting for the 
Quaternary Ice-Age as a whole by the latter's theory of 
continental displacements (Chapter XIV.), bring in astro- 
nomical causes to account for the succession of glacial and 
interglacial periods. Like Spitaler, they emphasise the 
importance of a low summer temperature, but they employ 
a different method, based on work by M. Milankovitch (i), 
in which the amount of radiation received at any point during 
the summer half-year is expressed as the equivalent latitude 
of that point with respect to present astronomical conditions. 
With perihelion in June and a great obliquity of the ecliptic, 
the radiation received by any point in the north temperate 


or Arctic zones will be greater than the present, so that the 
equivalent latitude will be lower. Thus, in 65 N., the 
summer radiation in 9,500 B.C. was as great as that now 
received in 60 20' N., while in 20,400 B.C. the summer radiation 
was as low as that now received in 68 N. It is pointed out 
that a decreased obliquity of the ecliptic must increase the 
temperature contrast between pole and equator in summer 
without making much difference in winter. Hence the 
atmospheric circulation in summer would be strengthened, 
giving an increased frequency and intensity of cyclones in 
the temperate belt, and the summer rainfall of Southern 
Europe would be greater. 

The plane of the ecliptic varies in a period of 40,400 years, 
while perihelion makes the circuit of the seasons in 20,700 years. 
In each hemisphere the coldest summers occur when the 
smallest obliquity corresponds with summer in aphelion 
during a period of maximum eccentricity. Hence the authors 
suppose that the glacial periods occur at maximum eccentricity, 
and consist of two periods of cold summers 40,000 years apart, 
which are united in one glaciation through the power of 
persistence of an ice-sheet. Any retreat stadia formed during 
the intervening period of warmer summers are masked by the 
readvance of the second phase, which starts from the remains 
of the first ice-sheet and therefore attains a greater develop- 
ment. The glaciations in the two hemispheres are therefore 
roughly synchronous, but a maximum in the Southern 
Hemisphere occurs about 10,000 years before or after the 
corresponding maximum in the Northern Hemisphere. On 
these lines the authors give in Table 3 a chronology of the 
Quaternary Ice-Age. 

This theory marks a great advance on Croll and Spitaler 
but the chronology is still a long way from fitting the geological 
evidence, ending the Wurm glaciation, for example, about 
66,000 B.C. and dating the post-glacial climatic optimum at 
9,100 B.C. instead of 5-3,000 B.C. when the astronomical 
climate differed from the present by an insignificant amount. 

The latest and most elaborate reconstruction of glacial 
history from astronomical data, also based on Milankovitch, 
is by F. E. Zeuner (6). Zeuner studies the glacial history of 
Europe in minute detail, including also the periglacial regions, 
from which he makes inferences as to the character of the 


Northern Hemisphere. Southern Hemisphere. 

(The figures are in thousands of years before the present.) 
Climatic Optimum, 9-1. 

Pre-Baltic Stadium, 33-30. 
Post-Wurm I., 110-103. 

Baltic Stadium, 25. 

Wurm II., 74-66. 
Wurm L, 118-110. 

RissIL, 193-183. 
Riss L, 236-225. 

Nameless, 1 305-302. 

Mindel II., 434-429. 
Mindel I., 478-470. 

Gunz II., 550-543. 
Gunz I., 592-585. 

Post-Riss II., 200-195. 
Post-Riss L, 226-218. 

About (442), 389, 350, 
312, 270. 

Post- Mindel I., 468-462. 
Pre-Gunz II., 560-554. 
Table 3. Koppen's interpretation of Quaternary sequence. 

climate. He follows Koppen and Wegener in expressing 
variation of summer radiation as equivalent latitude. He 
shows that the variation of the present snow-line with latitude 
closely follows the variation of radiation received in the 
summer half-year. He argues that with changing astro- 
nomical conditions a rise of the winter temperature increases 
the snowfall and the corresponding fall of summer temperature 
enables the snow to persist through the year. He also follows 
up various secondary effects of glaciation such as the reflecting 
power of the snow surface, shift of tracks of depressions, and 
the periglacial belt of east winds, as well as the effect of changes 
of sea level caused by the locking up of great quantities of 
water in the ice-sheets, and the delayed effect of the weight 
of the ice in depressing the land. 

The result is a very detailed scheme of changes of both 
climate and sea-level which fits in well with the most recent 
geological interpretations, as exemplified by the curve of 

1 H. Gams and R. Nordhagen consider that between the Mindel and Riss 
glaciations in the Alps, but nearer the latter than the former, there was an 
additional glaciation, which they term the Muhlbergian. 



equivalent latitude of 65 N. His dating is briefly as follows 
(dates in thousands of years before present day) : 


Late Glaciation III. 





Antepenultimate 1 1 . 
Glaciation I. 

Early Glaciation II. 


. 7 2 




Penck and Bruckner. 
Wurm 40-18. 






Table 4. Zeuner's interpretation of Quaternary sequence. 

On Zeuner' s view the long interglacial between the Ante- 
penultimate and Penultimate Glaciations (or Mindel-Riss) 
was on the whole about 4 F. warmer than the present, with 
a rather oceanic climate, but it was interrupted by a minor 
cold period (see footnote to Table 3). 

Zeuner recognises that the astronomical theory does not 
account for the Quaternary Ice- Age as a whole, only for the 
details within the period. His scheme comes much nearer to 
the geological dating than do those of Spitaler and Koppen 
and Wegener. The astronomical dates still tend to be earlier 
than those favoured by the geologists, but this may reasonably 
be attributed to the natural lag in the accumulation and 
disappearance of great ice-sheets, which Zeuner considers 
may have amounted to some thousands of years. 

Whatever we may think of the parallelism between 
Milankovitch's curve and the succession of glacial stages, it 
is clear that astronomical changes, and especially changes in 
the eccentricity of the earth's orbit, must have had quite 
appreciable effects on the climates of the past. At periods 
of maximum eccentricity the hemisphere with winter in 
aphelion must, other things being equal, have more pronounced 
seasons than that with winter in perihelion, and these pro- 
nounced seasons must have increased the strength of monsoons 
and other seasonal changes. 


According to W. H. Bradley (7) the Eocene of Colorado, 
Utah and Wyoming includes beds with annual layers covering 
a duration of five to eight million years, which show periodi- 
cities of 1 1\ years (sunspot cycle), 23 years, 50 years and about 
21,000 years, the latter presumably representing the cyclic 
changes of eccentricity of the earth's orbit and the precession 
of the equinoxes. Alternations of layers in the Cretaceous of 
U.S.A. also suggest a cycle which is estimated as about 
21,000 years, but there are no annual layers. 

It is possible that the coal seams in the Upper Carboniferous 
represent periods of great eccentricity with winter in perihelion 
in the Northern Hemisphere and a small obliquity of the 
ecliptic giving a climate in the Northern Hemisphere with 
little annual range of temperature, and consequently no 
annual rings of growth in the woody stems. In that case the 
intervening beds of sandstone and clay would represent the 
other halves of the periods in which winter in the Northern 
Hemisphere was in aphelion and the climate consequently 
more extreme, and the whole succession from the base of one 
coal seam to the base of the next would represent a period of 
21,000 years. Nearer the present, the alternation of brown 
coal and bauxite beds in the Tertiary of South-eastern Europe 
may be due, as suggested by Koppen and Wegener, to a 
similar alternation of equable and extreme astronomical 
climates. Similarly, as suggested by Dacque (8), the extreme 
climate of the Old Red Sandstone may be due to the deposits 
Raving been formed during a period of great eccentricity, 
but in the absence of absolute dating of these deposits, both 
these suggestions must remain speculative. There is more 
support for the supposition that during the closing stages of 
the retreat of the Quaternary ice-sheet of Scandinavia, the 
extreme " continentality " of the climate was due partly to 
the greater obliquity. The maximum obliquity of 8500 to 
7500 B.C. (according to the work of Stockwell) would have 
had the effect of lowering the winter temperature and raising 
the summer temperature by about i F. Even in this example, 
however, the changes due to purely geographical causes were 
probably much greater than those due to the increased 
obliquity, and the astronomical effect by itself would have 
been hard to distinguish. 



(1) MILANKOVITGH, M. " Th^orie math^matique des phnomenes thermiques 

produits par la radiation solaire." Paris, 1920. 

(2) GROLL,J. " Climate and time in their geological relations." London, 1875. 

(3) SPITALER, R. " Das Klima des Eiszeitalters." Prag, 1921 (Lithographed). 

(4) MURPHY, J. J. " Glacial climate and polar ice-cap." London, Q,. jf. 

GeoL Soc., 32, 1876, p. 400. 

(5) KOPPEN, W., and A. WEGENER. " Die Klimate der geologischen Vorzeit." 

Berlin, 1924. 

(6) ZEUNER, F. E. " The Pleistocene period ; its climate, chronology and 

faunal successions." London, Ray Soc., 1945. 

(7) BRADLEY, W. H. " The varves and climate of the Green River epoch." 

U.S. Geol. Surv., Prof. Paper, 158, 1929, p. 87. 

(8) DACQUE, E. " Grundlagen und Methoden der Palaeogeographie." Jena, 



THE mean temperature of the earth is determined 
by the balance between the radiation received from 
the sun and that given out by the earth. The solar 
radiation entering the earth's atmosphere undergoes various 
transformations before it finally leaves the atmosphere again 
as radiation to space. These changes have been set out very 
clearly in a diagram by W. H. Dines (i) (Fig. n, reproduced 
by permission of the Royal Meteorological Society). In this 
diagram, A is the radiation reaching the limit of the earth's 
atmosphere from the sun. The numerical value of A 
(measured in calories per square centimetre per day) is about 
720. Part of this radiation is reflected from the surfaces of 
clouds, snowfields, and to a lesser degree from all parts of the 
earth's surface, both land and water, without undergoing any 
change. Another part of the solar beam is scattered by the 
molecules of the gases and the particles of dust in the atmos- 
phere ; some of this scattered radiation ultimately reaches the 
surface of the earth, but some of it is entirely lost to the earth. 
The amount lost by reflection and scattering is represented by 
D, which has a value of about 320. [The numerical values 
are taken partly from a later source (2).] Another part of 
the radiation, G, is absorbed by the air, and the remainder, 
B, is absorbed by the earth. The average value of B is about 
350, leaving only 50 for C. 

The surface of the earth is losing heat in three ways. An 
amount, G, which depends on the temperature, is radiated 
outwards ; of this a small part, M, is reflected back to the 
earth, mainly from the under surfaces of clouds, without 
undergoing any change ; another part, H, is absorbed by 
the air, and the remainder, K, is lost to space. Another 
part of the surface heat is transferred to the atmosphere by 
evaporation of moisture from the surface, the heat of evapora- 
tion being given up to the air when the moisture is again 


condensed, and a third part is communicated from the earth 
to the air by conduction and carried upwards by convection, 
but this is partly balanced by the reverse process, conduction 
from the air to the earth, and the net result is probably small. 
The heat transferred by evaporation and by the net conduction 
together is indicated by L. 

The air is constantly receiving heat, C, from the sun, H 
and L from the earth. This heat is in turn radiated by the 
air, part E going back to the earth and part F to space. Thus 
the incoming radiation A is finally given out again in three 
forms, D by direct reflection, K by radiation from the earth, 
and F by radiation from the air, and since, practically speaking, 





k ) 


Outer Limit of 





A,r t 


*~ H 






XT _ 

) ( 

)G , 


Fig. 1 1 . Heat exchange of atmosphere. 

there is no gain or loss of heat from one year to the next, 

The value of G, the radiation from the earth's surface, 
is related to the temperature in accordance with the well- 
known Stefan-Boltzmann law, which states that the radiation 
from a black body is proportional to the fourth power of its 
absolute temperature. The earth's surface is not quite a 
black body, but the radiation must be greater the higher the 
temperature. G and L are together equal to B+E + M, hence 
the temperature of the earth's surface must be determined by 
the sum of these three quantities (radiation from the sun 
reaching the surface, radiation from the air to the earth, 
and terrestrial radiation reflected back to the earth), less the 
heat L lost by evaporation and conduction. A change in 
any of these quantities will therefore bring about a change 
in the mean temperature of the earth. W. H. Dines (i) and 


(2) assigns numerical values to all the quantities A to M ; 
some of these values are only rough estimates, but they will 
serve for a discussion of the possibility of appreciable climatic 
changes being brought about by variations in their amount. 

The amount of solar radiation absorbed by the air is 
probably small ; Dines gives it a value of 50 calories, or 
one-fourteenth of the whole solar radiation, so that for the 
present we can ignore it. Variations in the value of A have 
already been discussed in Chapter IV., and here A is con- 
sidered as a constant. Variations of B, the solar radiation 
reaching the earth's surface unchanged, therefore depend 
chiefly on variations of D, the solar radiation reflected back 
to space or lost by scattering ; Dines gives a value of 320 for 
D. The chief reflecting surfaces are clouds and snowfields, 
and the loss by scattering may be greatly increased by the 
presence of dust. Clouds are very efficient reflectors, as is 
shown by the intense brightness of cumulus clouds in sunlight ; 
A. Angstrom estimates that clouds reflect 75 per cent, of the 
solar radiation falling on them, while L. B. Aldrich (3) puts 
the figure at 78 per cent, and, calculates that on the average 
slightly over 40 per cent, of the sun's radiation is lost to the 
earth by reflection from clouds. Hence we should expect an 
increase of cloudiness to lower the average temperature. 
At present the average cloudiness over the whole world is 
just over five-tenths ; if it increased to six-tenths without 
any corresponding increase in the amount of water held in 
the atmosphere as vapour (clouds are water droplets or ice 
crystals, not water vapour, so that the supposition is con- 
ceivable), the mean temperature would be considerably 
lower than it is now. This effect of cloudiness is of great 
importance ; it is discussed in greater detail in the next 
chapter. If the increase of cloudiness were accompanied 
by an increase in the amount of water vapour, the fall of 
temperature would be less, as will be seen. 

An increase in the area covered by snow and ice would 
increase the heat lost by reflection, and in this case there are 
no compensating circumstances. Large areas of snow and 
ice usually have clear skies and dry air above them, and there 
is nothing to check the wastage of o solar heat. If we assume 
on the basis of some work by A. Angstrom (4) that over an 
ice-sheet about four-fifths of the solar radiation is reflected 


back to space, compared with one-fifth in an unglaciated 
region, the lost solar radiation in temperate latitudes would 
be sufficient to melt more than 30 feet of ice in the course of 
a year. This must have been a powerful factor in increasing 
the rigour of the great ice-ages in the glaciated regions, and 
in facilitating the extension of the ice-sheets. According to 
E. Antevs (5), the ice-covered area at the maximum of 
glaciation was about 13 million square miles, compared with 
6 million square miles at present. The increase of 7 million 
square miles represents 3 5 per cent, of the earth's surface. 
Allowing for the fact that glaciated regions are not entirely 
cloudless, we may estimate the increased loss of solar radiation 
by reflection from the ice-sheets as at least 2| per cent, of that 
reaching the whole earth. W. Wundt (6), allowing for the 
snowfall and drift ice of the peripheral regions, arrives at a 
figure of 3 per cent. The effect would be to lower the 
temperature of the whole earth by at least 4 F. (Wundt gives 

7 F.). 

A variable which may be more generally effective is the 
volcanic dust in the atmosphere. In the years following 
great volcanic eruptions of the explosive type, such as those 
of Krakatoa (1883), Santa Maria and Pelee (1902), Colima 
(1903), and Katmai, Alaska (1912), the solar radiation 
reaching the earth's surface (Dines 5 quantity B) may be 15 
or 20 per cent. (45 to 60 calories) below the normal value. 
This radiation is not all lost to the earth ; part of it goes to 
warming the upper air and therefore increases the value of 
E, but we shall see later that the net effect is an appreciable 
lowering of temperature, and that volcanic dust has possibly 
played a part in causing ice-ages. 

The quantity M, terrestrial radiation reflected back to 
the earth, is relatively small (60 calories) on Dines' calculation, 
and as its variations depend on some of the factors, cloudiness 
and dust, which control D, it need not be considered further 
here. That leaves for discussion E, the radiation from the 
air to the earth, and this quantity is not only large (540 
calories) but probably very variable. It depends on several 
factors, but by far the most important is H, the terrestrial 
radiation absorbed by the atmosphere, to which Dines assigns 
a value of 600 calories per square centimetre per day. The 
action of the atmosphere in raising the temperature of the 


earth by absorbing its radiation is similar to the action of the 
glass roof in a greenhouse. In the atmosphere the place of 
the glass is taken by certain gases, notably water vapour, 
ozone, and carbon dioxide. W. H. Dines points out (2) that 
the present mean temperature of the earth's surface, 288 A. 
(59 F-)> * s a bout 4 A. (7 F.) above the mean temperature 
which would prevail if the atmosphere had no power of 
absorption and if there were no reflection or scattering. In 
addition to raising the mean temperature over the earth as a 
whole, atmospheric absorption, in conjunction with the 
mobility of the air, has an important effect in bringing about 
a greater approach to uniformity in the temperatures of 
different latitudes than would exist on a dry earth. The 
importance of this effect in controlling variations of climate 
has been emphasised by Sir George Simpson. 

Water vapour has a very high coefficient of absorption of 
radiation in the wave-lengths chiefly emitted by the earth. 
Moist air is warmed by radiation from the earth and 
immediately returns to the surface part of the heat gained in 
this way. Hence on a clear night the surface loses heat more 
rapidly if the air is dry than if it is moist, and the amount of 
water vapour in the lowest layers is one of the most important 
elements in calculations for forecasting the occurrence of 
frosts. Except over cold dry continental regions, however, 
there is always enough water vapour present in the whole 
thickness of the atmosphere to absorb practically the whole of 
the radiation in most parts of the spectrum of terrestrial 
radiation. An increase in the amount of water vapour 
would therefore have little effect on the proportion of 
terrestrial radiation transmitted unchanged to space. It 
would, however, have some effect on the vertical distribution 
of temperature in the atmosphere. Consider the atmosphere 
as divided into a number of concentric shells, and suppose for 
a moment that each of these shells absorbs all the radiatiom 
reaching it from the shell on either side, while there is a fall 
of temperature from the inner to the outer shells. Then it 
will be seen that the earth's surface is receiving radiation only 
from the innermost shell, while only the outermost shell is 
radiating to space. Hence the radiation from the air to the 
earth is greater than the radiation from the air to space. 
Although these conditions are not fully realised in the earth's 


atmosphere, they are sufficiently near the truth for the above 
conclusion to hold. An increase of water vapour, by increasing 
the completeness of absorption in each of our hypothetical 
shells, would make this difference somewhat greater, and 
so raise the temperature of the earth's surface, while it 
would increase the temperature of the air at a height of 
four or five miles more than that at the surface, and so lessen 
the decrease of temperature with height. 

More aqueous vapour in the air may be due either to 
greater evaporation at the same temperature or to a higher 
initial temperature. Greater evaporation may be due to a 
greater expanse of sea in low latitudes, to a greater average 
wind velocity, or to a greater vertical interchange of air. 
In any case more heat is taken from the surface and carried 
to the level of condensation, whence it is partly radiated to 
space. The result is an increase of L, the heat lost to the 
earth's surface otherwise than by radiation, and it therefore 
causes a decrease of the earth's radiation G, i.e., a fall of the 
surface temperature, in spite of the greater humidity. On 
the other hand, if the mean temperature of the air rises from 
some cause unconnected with the water vapour content, the 
latter will automatically increase, since warm air has a greater 
capacity for water vapour than cold air has. If all other 
conditions of land and sea distribution, relief, etc., remained 
unchanged, an increase in the water-vapour content due to 
increased solar radiation would be accompanied by an increase 
in cloudiness and precipitation as well as by a rise of tem- 
perature ; this is the basis of Sir George Simpson's theory of 
glaciation described on pages 91 to 96. If, however, an 
increase in the water vapour content is brought about by a 
large increase in the area of the oceans, accompanied by a 
decrease in the average height of the land, the solar radiation 
remaining unchanged, there will not necessarily be an increase 
of cloud and precipitation. 

The most important way in which cloud is formed, and 
practically the only way in which an appreciable fall of rain 
can be produced, is by the ascent of moist air, which is cooled 
by expansion until condensation takes place. Except in 
limited regions of high relief, the air must become unstable 
before is can be forced to rise. The stability of a column of air 
depends on the temperature gradient ; the less rapid the fall 


of temperature with height, the more stable is the air. We 
have seen that an increase in the water vapour, other things 
remaining unchanged, increases the absorption at high levels 
in the atmosphere, and therefore decreases the fall of tem- 
perature with height and makes the air more stable. The 
ascent of air, both in the great cyclonic storms and " barometric 
depressions " and in the smaller thunderstorms and local 
showers, becomes less frequent and extensive, and there may 
be an actual decrease in the cloudiness. These conditions are 
illustrated in the trade wind belts. The air crosses large 
stretches of ocean ; it speedily becomes nearly saturated and 
takes up further moisture very slowly, yet owing to the stability 
of the conditions there is very little ascent of air, the amount 
of cloud is small, and rainfall is very scanty. With a generally 
higher temperature over the earth, these conditions would 
extend to higher latitudes. During the hot summer of 1921 
the air over the British Isles contained more water vapour than 
during the cool summer of 1924, but in 1921 the air was stable 
and there was little cloud and rainfall, while in 1924 the air 
was unstable and there was much cloud and a rainfall above 

The development of barometric depressions and conse- 
quently the formation of cloud are facilitated by the presence 
of horizontal temperature differences in masses of air which 
are moving relatively to each other. A more uniform dis- 
tribution of temperature over the earth, such as prevailed 
during the warm periods, would therefore largely remove one 
important source of cloudiness at the present day. Thus it 
is possible that a general elevation of temperature over the 
globe could increase the amount of water vapour in the air 
without increasing the cloudiness ; the result would be a 
further rise of temperature, intensifying the original increase. 

In certain theories of climatic change (See Appendix II., 
Section IX.) extraordinary importance is attached to variations 
in the amount of carbon dioxide in the atmosphere. v Periods 
during which the atmosphere was rich in this gas are con- 
sidered to have been uniformly warm, those in which it was 
poor to have been cold or even glacial periods. Chamberlin 
has devoted great ingenuity to a discussion of the probable 
variations in the quantity of carbon dioxide in the atmosphere 
during different geological periods, and though his conclusions 


are probably somewhat exaggerated, there appears to be little 
doubt that the variations have been considerable. Recent 
physical researches have shown, however, that the part of 
the terrestrial radiation which is taken up by carbon dioxide 
is almost completely absorbed by water vapour, and no 
increase in the amount of the former gas could increase the 
total absorption appreciably. W. J. Humphreys pointed out 
(7, p. 567) that the only way in which an increase of carbon 
dioxide could affect the temperature would be by absorption 
at high levels in the atmosphere where water vapour is nearly 
absent. As explained in the case of water vapour, this would 
increase slightly the proportion of radiation from the air which 
is directed towards the earth, and decrease that which is 
directed towards space, and in this way, the mean temperature 
of the earth may have been modified to the extent of a few 
tenths of a degree. 

In 1939, however, the question was taken up again by 
G. S. Callendar (8) who relates the cold of the Permian 
to the exhaustion of carbon dioxide by the Carboniferous 
forests. During the Mesozoic the relatively small develop- 
ment of plant life allowed the amount of CO 2 , steadily 
replenished by the animal life of the seas, to increase again, 
but a great deal was locked up in Tertiary lignite formation, 
especially in western North America, and this may have 
brought about a progressive cooling which ended in the 
Quaternary Ice-Age. This theory cannot account for the 
oscillations of the individual glaciations, the time-scale of 
which is too short. Callendar ends by pointing out that the 
great coal consumption in the twentieth century has raised 
the amount of CO 2 in the atmosphere from -028 per cent, 
about 1900 to -030 per cent, in the i93o's, and that this 
increase has been accompanied by a small but steady rise in 
the mean temperature of the colder regions of the earth. 
This argument has rather broken down in the last few years, 
however, for the rise of temperature seems to have reached 
its crest and to have given place to a fall. The possibility 
that changes in the amount of CO 2 have been responsible for 
some small part of the climatic changes of geological time 
seems to remain open however. 

The idea that volcanic dust may have important climatic 
effects is very old. In 1784, B. Franklin suggested that the 


hard winter of 1783-84 was due to the great quantities of 
dust in the air, and that the source of this dust might be 
either the destruction of meteorites or the great volcanic 
eruptions in Iceland. Other references to this possibility 
appeared from time to time, but it was not until the great 
diminution of radiation received at the surface after the 
eruption of Katmai in 1912 was noted that the subject received 
exhaustive discussion by Abbot and Fowle (9), and W. J. 
Humphreys (10, also 7, pp. 569-603). This discussion has 
shown that the effect of dust is due, not to absorption, but 
to the scattering and reflection of the solar radiation to which 
it gives rise. When a volcanic dust cloud is thrown to a great 
height in the air, as in the eruption of Krakatoa, it takes from 
one to three years to settle. Humphreys calculated the average 
diameter of the particles from the optical effects, and found 
that it was i -85 microns (-00185 mm.). This is greater than 
the wave-length of solar radiation in the region of maximum 
intensity, and the particles therefore greatly interfere with 
the passage of the solar radiation. On the other hand, the 
wave-length of terrestrial radiation is six or seven times the 
diameter of the particles, and the terrestrial radiation passes 
through dusty air with little loss. The action of dust in the 
two cases may be compared to that of a number of wooden 
balls floating in a pond. Against a barrier of such balls 
small ripples break up and are lost, but larger waves to which 
the balls can rise and fall are hardly affected. Humphreys 
calculates that the reduction by volcanic dust of solar and of 
terrestrial radiation is in the ratio 30 to i. Observations in 
1912 showed that the Katmai dust reduced the solar radiation 
reaching the earth by about 20 per cent., which, if maintained 
through a long period of years, would lower the mean tem- 
perature of the earth by about 10 F., an amount quite 
sufficient to initiate an ice-age. The compensating processes 
reflection of terrestrial radiation, and radiation from the 
dust itself are negligible in comparison. During the past 
1 60 years the average temperature of the earth has been 
lowered by volcanic dust possibly as much as i F. The 
most notable cold years since the beginning of the eighteenth 
century have all followed great volcanic eruptions, especially 
1784-86, following the eruption of Asama (Japan) in 1783 ; 
1816 (" he year without a summer"), following especially 


the eruption of Tomboro (Sumbawa) in 1815 ; 1884-86, 
following Krakatoa in 1883 ; and 1912-13, following Katmai 
in 1912. H. Arctowski (n), who has studied the temperature 
variations in great detail in connexion with the solar control 
of terrestrial temperatures, admits the great influence of the 
Krakatoa eruption, but considers that the effects of the 
eruptions of 1902 and 1912 on temperature were very small. 
The direct effect of volcanic dust on temperature is felt over 
the whole globe (according to Arctowski only over the hemi- 
sphere in which the eruption occurred) ; it is not localised 
in the regions which were glaciated during the Quaternary 
Ice-Age, but in addition to, or in consequence of, the direct 
effect, there may be a secondary effect on the centres of high 
and low pressure. A. Defant (12), from a study of the strength 
of the atmospheric circulation during the two years following 
each of the four great explosive volcanic eruptions of 1883 
(Krakatoa), 1886 (Tarawera), 1888 (Bandai San) and 1902 
(West Indies), considered that the result of each eruption 
was a strengthening of the atmospheric circulation for the 
next two years. C. E. P. Brooks and T. M. Hunt, however, 
(13), taking into account also eruptions in 1875, *9 12 an d 1914* 
found that the strengthening of the circulation was much 
more transient, persisting for only six months. We are left 
therefore with a general cooling as the only appreciable 
climatic effect of volcanic dust. 

It may be remarked here that a large increase in the number 
of meteors entering the earth's atmosphere would have a 
similar effect to that of volcanic dust, but so far as I know, 
there is no evidence of such an increase during geological time. 

Volcanic dust appears to be a possible explanation of 
climatic periods colder than the present. The variation of 
this element in geological time is not yet well known ; 
Humphreys calculates that the amount of dust required to 
maintain an ice-age would amount to a layer only one-fiftieth 
of an inch thick in 100,000 years, so that it would hardly be 
noticeable in the sedimentary rocks. We should expect 
volcanic activity to be greatest during the great periods of 
earth-movement and mountain-building, when the continents 
were highly emergent and the land and sea distribution 
favourable for glaciation. The explosive stage of activity 
generally comes later in the life-history of a volcano than the 


stage of fluid eruptions, so that the maximum amount of dust 
might be delayed for some time after the continents first 
became highly emergent. In the Quaternary, at least, there 
was in fact an appreciable lag between the first great emergence 
of North America and Scandinavia and the beginning of 
widespread glaciation, and this difficulty is met by the 
hypothesis that the final lowering of temperature necessary 
for ice-formation was given by the occurrence of widespread 
explosive eruptions. On the other hand, the complete 
absence of volcanic dust would not raise the mean temperature 
more than about i F., which is inadequate to account for the 
temperature of the warm geological epochs. 

E. and A. Harle (14) attributed the general warmth of 
the Mesozoic period to the presence of a much denser atmos- 
phere than now exists on the earth, which helped to conserve 
the heat of the sun and so raised the general temperature. 
They base their argument on the existence of great flying 
reptiles in the Mesozoic, which they consider were too heavy 
to have flown in the present atmosphere, but as we are ignorant 
of the muscular development of these reptiles, this argument 
carries no weight. From the physical side there seems to be 
no reason why the earth should be losing its atmosphere, 
and the variations of climate during geological time are very 
far from suggesting the action of any irreversible force. 


(1) DINES, W. H. " The heat balance of the atmosphere." London, Q,. J. R. 

Meteor. Soc., 43, 1917, p. 151. 

(2) " Dictionary of Applied Physics," edited by Sir RICHARD GLAZEBROOK, 

vol. iii., London, 1923. Article, " Radiation," by W. H. DINES. 

(3) ALDRICH, L. B. " The reflecting power of clouds." Washington, D.C., 

Ann. Astrophys. Obs., Smithsonian Inst., 4, p. 375. 

(4) ANGSTROM, A. " The albedo of various surfaces of ground." Stockholm, 

Geogr. Ann., 7, 1925, p. 323. 

(5) ANTEVS, E. " The last glaciation." New York, Amer. Geogr. Soc., 

Research Series, no. 17, 1928. 

(6) WUNDT, W. " Anderungen der Erdalbedo wahrend der Eiszeit." Met. 

#., Braunschweig, 50, 1933, p. 241. 

(7) HUMPHREYS, W. J. " Physics of the air." 3 ed. London (McGraw-Hill), 


(8) CALLENDAR, G. S. " The composition of the atmosphere through the 

ages." Meteor. Mag., London, 74, 1939, p. 33. 

(9) ABBOT, C. G., and F. E. FOWLE. " Volcanoes and climate." Washington, 

D.G., Ann. Astrophys. Obs. Smithsonian Inst., 3, 1913, and Smithsonian Misc. 
Coll., 60, 1913, no. 29. 

( 10) HUMPHREYS, W. J. " Volcanic dust and other factors in the production of 
climatic changes and their possible relation to ice-ages." Philadelphia, 
J. Frankl. Inst., 176, 1913, p. 131. 


(u) ARCTOWSKI, H. "Volcanic dust veils and climatic variations." New 
York, Annals JV. Y. Acad. Sci., 26, 1915, p. 149. 

(12) DEFANT, A. " Die Schwankungen der atmospha rischen Zirkulation u'ber 

deni nord-atlantischen Ozean im 25-jahrigen Zeitraum, 1881-1905." 
Stockholm, Geogr. Ann., 6, 1924, p. 13. 

(13) BROOKS, C. E. P., and T. M. HUNT. " The influence of explosive volcanic 

eruptions on the subsequent pressure distribution over western Europe." 
Meteor. Mag., London, 64, 1929, p. 226. 

(14) HARLE, E., and A. HARLE. " Le vol de grands reptiles et insectes disparus 

semble indiquer un pression atmosphe"rique eleve"e." Paris, Bull. Soc. 
Geol. France, n, 1911, p. 118. 



IN the last chapter we saw that one of the most potent 
factors in modifying the distribution of solar radiation 
over the surface of the earth was the reflection from the 
upper surfaces of clouds. For this reason, cloudiness must 
obviously be of great importance as a factor in the mean 
temperature of any region. In Africa, for example, the highest 
mean annual temperatures are found, not over the equator 
where the solar radiation at the limit of the earth's atmosphere 
is greatest, but over the deserts to the north and south of the 
equator where the skies are clearest. The mean annual 
temperature (corrected to sea-level) over the equator in the 
interior of Africa is 82 F., while over the Sahara in latitude 
20 N. it is 89 F. In spite of the lower mean altitude of the 
sun, the small degree of cloudiness over the Sahara (2-4 tenths 
of sky covered in latitude 20, compared with 5 5 tenths at 
the equator) is sufficient to make the former 6 to 7 F. warmer 
than the latter. It is worth while to investigate this effect a 
little further, and attempt to make some estimate of the change 
of mean temperature which would result from small variations 
in the mean cloudiness, all other factors being regarded as 

We have seen that the mean temperature of the earth's 
surface is closely related to the amount of radiation which 
it gives out, and that the latter is equal to the amount of 
radiation which it receives, minus the loss of heat by 
evaporation and convection. If the earth were a perfect 
radiator, and if its temperature were the same in all parts 
of the surface, the fourth power of its temperature would 
be proportional to the amount of heat which it radiates 
outward. Although neither of these conditions is quite 
fulfilled, it is obvious that the higher the radiation the greater 
the mean temperature, and vice versa. We have also seen 
that the incoming radiation from the sun undergoes a number 


of changes, as a net result of which the earth's mean tem- 
perature is higher than it would be if there were no atmosphere. 
These changes are very complex, but we can say that an 
increase in the radiation received from the sun would increase 
all the quantities concerned. The radiation reaching the 
earth's surface would be greater, the return radiation from 
the earth to the atmosphere would be greater, consequently 
the amount of terrestrial radiation absorbed by the atmosphere 
would be increased, which would in turn increase the radiation 
from air to earth, and so on. An increase of 10 per cent, in the 
radiation received from the sun would therefore result in an 
increase of not exactly 10 per cent., but something of the order 
of 10 per cent., say, between 5 and 15 per cent., in the radiation 
from the earth's surface. This result would be independent 
of the local modifications in the distribution of temperature 
due to the readjustment of the wind systems, and also supposes 
that there is no change in the constitution of the atmosphere 
such as the formation of ozone in the stratosphere. 

Clouds being effective reflectors of solar radiation, the 
presence of clouds means a loss to the earth of a certain 
amount of energy. It is true that clouds also reflect a part 
of the terrestrial radiation back to the earth, and that this 
partly compensates for the lost solar radiation, but the com- 
pensation is not nearly complete. For one thing, it seems 
probable that the percentage of the long-wave terrestrial 
radiation reflected from the under surfaces of clouds is smaller 
than the percentage of the short-wave solar radiation reflected 
from their upper surfaces ; the action of water droplets in 
this respect is probably similar in kind, though perhaps not 
equal in degree to the action of volcanic dust particles described 
on page 118. The temperatures of the surface and of the 
atmosphere are so intimately connected that it is impossible 
to conceive of a great change in one without a change in the 
other. Hence we may say that it is immaterial whether the 
radiation is lost owing to a decrease in the amount of heat 
radiated by the sun or a decrease in the amount which 
penetrates the mantle of clouds. 

A. Angstrom ( i ) estimated the reflection of solar radiation 
falling on clouds as 75 per cent. This figure probably applies 
to rather dense heavy clouds ; more recently B. Haurwitz (2) 
investigated the insolation received at Blue Hill, Mass., with 


given cloud amount, cloud density and elevation of the sun. 
Cloud density is expressed on a scale of o (very thin) to 4 (very 
dense cloud), and is as important as cloud amount in deter- 
mining the insolation, but as we have no means of estimating 
variations of cloud density during geological time, it seems 
best to consider the results with the average density of 2 6 
found by Haurwitz. He gives a table of annual insolation 
with various cloud amounts and cloud densities as a percentage 
of the insolation with cloudless skies. Interpolating for 
density 2-6, we have approximately : 

Cloud amount (tenths) . o 1-3 4-7 8-9 10 
Insolation, per cent. .100 93 82 68 41 

At present the average cloudiness over the earth is 5 4 tenths, 
giving an average of 82 per cent, of the radiation received with 
a cloudless sky. An increase of one tenth in the cloud amount 
would reduce this figure to 78 per cent., and a decrease of 
one tenth of cloud would raise it to 85^ per cent. Let us 
make the simple assumption that the radiation from the earth's 
surface, and therefore the fourth power of the mean temperature 
(absolute degrees) are proportional to the amount of in- 
solation. Then, starting with a figure of 59 F. or 288 A. 
for the mean temperature of the earth's surface at present, we 
have the following results : 

4-4 5-4 6-4 

Cloudiness (tenths of sky) 
Insolation (per cent, of cloudless 

sky) 90 86 82 

Mean temperature, A. . . .291 288 284 
Mean temperature, F. . . . 65 59 53 

These figures show that a decrease in the mean cloudiness by 
only one tenth of the sky would result in an increase of the 
mean temperature by as much as 6 F., while an increase of 
cloudiness by the same amount would lower the mean 
temperature by 6 F. 

It may be noted here that large ice-sheets, by deflecting 
depressions into lower latitudes, would decrease the extent and 
permanency of the sub-tropical anticyclones, especially over 
the oceans. This in turn would lead to an increase in the 
cloudiness of these areas, and would result in an appreciable 


loss of heat to the world. This may have been a contributory 
cause of the rapid expansion of the ice-sheets. 

We saw in the Introduction that one of the salient features 
of the " warm " periods was their dryness. From this we 
naturally infer that they were favoured by unusually clear 
skies, and this gives us an explanation of their warmth or 
rather, it pushes the explanation one step farther back, for 
we still have to account for their clear skies. 

Up to the present we have been assuming, for the purposes 
of the argument, that the cloudiness of the sky is the same in 
all latitudes. That, of course, does not correctly represent 
the conditions prevailing at present. We have a belt of cloudy 
skies near the equator (about 5! tenths), then two belts of 
clear sky along latitudes 20 to 30 in each hemisphere, 
with an average cloudiness of 4 to 5 tenths, decreasing to 
less than 2 tenths in the centres of the great deserts, and 
outside these clear belts cloudiness increasing again to 6| or 
7 tenths in cold temperate and sub-polar regions. This 
distribution has a considerable effect on the mean tempera- 
ture of the earth, for two reasons. First, owing to the greater 
average elevation of the jsun, the radiation received (at the 
limit of the atmosphere) is much greater in low latitudes 
than it is near the poles. The radiation received on a hori- 
zontal surface of one square centimetre area in the course 
of a year on the Arctic Circle is about half that received on 
the equator. Hence a cloud on the equator reflects back to 
space about twice as much radiation as a similar cloud in 
latitude 66. The same average cloudiness for the whole 
earth would result in a much higher mean temperature if 
the clouds were concentrated in high latitudes than if they 
were concentrated in low latitudes. 

Secondly, although clouds are much more effective as 
reflectors of solar radiation than as reflectors of the long- 
wave terrestrial radiation, they are not without a certain 
effect on the latter. The figure given by W. H. Dines 
(Chapter VI., i) is equivalent to a reflection of 8 per cent, 
of the terrestrial radiation compared with 60 per cent, of 
the solar radiation according to Haurwitz or 75 per cent, 
according to A. Angstrom. There is a large transference of 
heat from low to high latitudes by winds and ocean currents, 
and the temperatures of the polar and sub-polar regions are 


raised considerably by the heat conveyed in this way. Hence 
the outgoing radiation from the earth's surface in high latitudes 
is much greater than the incoming radiation ; let us suppose 
that in some particular locality it is ten times as great, and that 
the sky is half covered by clouds. Then it is easily seen that 
the gain of heat by the reflection of terrestrial radiation back 
to the earth exceeds the loss due to reflection of solar radiation 
back to space. Beyond the latitudes 67 N. and S., during the 
polar night, the incoming radiation from the sun is zero, and 
the effect is pure gain. Thus in high latitudes in winter, 
cloudy skies are actually effective in raising the mean 

The qualification " in winter 5> suggests that in addition 
to the mean cloudiness and its distribution according to 
latitude, the seasonal variation is important. A locality in 
which the sky is generally overcast in winter and clear in 
suirtmer would have a higher mean temperature than another 
place in the same latitude subjected to similar conditions, but 
with its skies clear in winter and overcast in summer. This 
is brought out very clearly by A. Angstrom (i) in an analysis 
of the annual variation of temperature at Stockholm in 
relation to the radiation. At Stockholm the mean annual 
cloudiness is 6-4 tenths, and it o varies from 5- 1 tenths in June 
to 7 9 tenths in December. Angstrom calculates that if the 
mean annual cloudiness remained at 6*4, but instead of being 
greater in winter than in summer was the same in all months 
of the year, Stockholm would be colder than it is at present 
in every month, the average difference being 2-2 F. 

The remains of desert deposits formed during the warm 
periods are mainly limited to middle and low latitudes, 
while in high latitudes we find the remains of a rich vegetation 
requiring a considerable rainfall. This distribution suggests 
that while the cloudiness was small between about 10 and 
55 latitude, it increased very rapidly beyond 55, thus giving 
the most favourable conditions for a high temperature in all 
parts of the world. 

Although the popular conception of a geological land- 
scape is a steaming jungle rather than, as it should be, an 
arid plain, variations of cloudiness have played compara- 
tively little part in theories of climatic change. Marsden 
Manson (3) has, however, seized on the dual role of clouds 


as reflectors alike of solar radiation and of terrestrial radiation, 
and has constructed an elaborate theory of geological climates 
on this basis, in conjunction with the gradual waning of the 
internal heat of the earth. He points out that the surface of 
the larger planets is covered by an unbroken layer of cloud, 
and assumes that this must have been the condition of the earth 
in past times, and in fact comparatively recently. Through 
this cloud canopy the sun's rays could not penetrate, and as a 
factor of climate the sun was almost inoperative. The earth 
was radiating more than it was absorbing, and the sources of 
this outgoing energy were the original supply of earth-heat 
and radio-active minerals. Owing to the poorly conducting 
crust, earth-heat was liberated, not in a steady stream, but 
in spasms during periods of volcanic action and crustal 

Let us start with one of these liberations of earth-heat. 
Within its protecting cloud canopy the surface, oceans and 
continents alike, was warm from equator to poles, but the land 
surfaces cooled more quickly than the heat-conserving oceans, 
and in due course, while the warm oceans were still supplying 
enough moisture to maintain the cloud canopy intact, the 
land surfaces began to freeze, and ice-sheets developed. 
Apart from some local glaciations in the centres of the larger 
continents this stage was first reached on a planetary scale 
in the Permo-Carboniferous period ; this glaciation coincided 
more or less with the present sub-tropical high-pressure belts, 
and the reason is stated to be that "cold an ti cyclonic winds" 
cooled the land most rapidly in those belts. The cooling of the 
oceans continued, and with decreasing evaporation a stage 
was reached in which these high-pressure areas, to-day 
possessing the clearest skies of the world, ceased to be mantled 
in clouds the sun broke through and deglaciation commenced. 

Now followed a period of dual control, solar energy 
prevailing near the equator, earth-heat towards the poles. 
In spite of fluctuations, the latter gradually diminished, and 
just before the Quaternary glaciation the polar oceans became 
cold for the first time. Then the second planetary glaciation 
occurred, centred in the cold temperate belts of greatest 
precipitation, at this time the only regions which were per- 
manently overcast. The cooling of the oceans continued, 
and evaporation ceased to supply enough moisture for even 


this limited cloud belt, the sun shone over the whole world, 
deglaciation again commenced and is still continuing. 

The theory is interesting, but there are some insuperable 
difficulties. With warm oceans and an unbroken cloud 
canopy, the land surfaces, unless at a great altitude, would 
not be likely to freeze ; the conditions are most nearly realised 
at present in the equatorial rain belt, in which the land is 
maintained at the same temperature as the neighbouring 
oceans. " Cold anticyclonic winds " presuppose cooling by 
radiation ; even if under world-wide isothermal conditions 
the pressure distribution could remain unaltered, which is 
highly improbable, we must suppose either that the anticyclone 
would break down the cloud canopy, in which case the 
tropical sun would certainly prevent glaciation, or that the 
clouds would remain in spite of the anticyclone, in which 
case the descending air would not be cold. Finally, the moist 
conditions supposed by Marsden Manson to have prevailed 
during the warm periods are in direct opposition to the dry 
conditions demonstrated by the geological evidence set out 
in the Introduction. 

L. J. Krige (4) suggested that increased cloudiness and 
precipitation would occur during periods of mountain building 
because of unusually high evaporation from ocean basins 
due to heat entering them from below. This type of suggestion 
is very difficult to discuss quantitatively but it seems that to 
make an appreciable difference to the amount of evaporation 
the quantity of earth-heat would have to be an improbably 
high multiple of the present supply. 


(1) ANGSTROM, A. " On radiation and climate." Stockholm, Geogr. Ann., 7, 

1925, p. 122. 

(2) HAURWITZ, B. " Insolation in relation to cloudiness and cloud density." 

jf. Met. Amer. Met. Soc., 2, 1945, p. 154. 

(3) MANSON, MARSDEN. " The evolution of climates." Baltimore, Md., 1922. 

(4) KRIGE, L. J. " Magmatic cycles, continental drift and ice-ages." GeoL\Soc. 

S. Africa, 1929. 



IN the last few chapters we have discussed the factors 
influencing the distribution of temperature, namely, winds 
and ocean currents, the heat received from the sun, the 
heat transmitted by the earth's atmosphere, and the reflection 
from cloud surfaces. We may divide these factors into those 
which are constant along a given line of latitude and give rise 
to the " solar climate," and those which vary from place to 
place even in the same latitude, and so give rise to the local 
distribution of climates. The latter may be termed the 
" geographical climate," since it depends mainly on the dis- 
tribution of land and sea and partly also on the relief of the 
land surface. Hence there is an intimate relationship between 
temperature and land and sea distribution, and we shall expect 
to find that the changing outlines of the geological continents 
have been reflected as changes in the distribution of 

If we look at a map of the mean annual temperature, we 
notice first of all that the isotherms run roughly parallel 
with the lines of latitude. There is a belt on both sides of 
the equator in which the temperature is above 80 F., 
extending from America across the Atlantic, Africa, India 
and the Indian Ocean, the East Indies, and part of the 
Western Pacific. This belt broadens out greatly over the 
continents and narrows over the oceans ; over the Eastern 
Pacific it thins out altogether. Between 30 N. and 30 S. 
the eastern sides of the continents are warmer than the 
western sides. Surrounding each pole is an irregular area 
over which the mean temperature is below freezing point ; 
the average position of the isotherm of 32 F. is north of 
75 N. in the Arctic, but about 60 S. in the Antarctic. In 
the Arctic the isotherm of 32 F. lies farther north over the 
oceans and the western coasts of the continents than it does 
over the central and eastern parts of the continents, that is, 
the land is generally colder than the sea. Hence in temperate 
















regions the isotherms are widely separated over the oceans 
and crowded together over the continents. The course of the 
warm Gulf Stream Drift is marked by poleward bends of the 
isotherms ; the course of the cold currents and the up-welling 
of cold water is marked by equatorward bends, especially 
along the California, Humboldt, and Benguela Currents. 

The charts for the extreme months January and July 
(Figs. 12 and 13) show that in temperate latitudes the sea 
is much warmer than the land in winter ; in summer the 
land is warmer, though not to the same extent. The larger 
the continent, the greater the depression of temperature in 
its centre, though this is controlled also by the roughness of 
the relief. The chart for January also shows the extraordinary 
effect which is exercised by the Gulf Stream Drift on the coast 
of Scotland and Norway ; in this month the Arctic Circle 
between longitude o and the Norwegian coast is actually 
more than 40 F. warmer than the average of the whole 
parallel, and 90 F. warmer than the " cold pole " of Siberia. 
The last point calling for special notice is the low temperature, 
in summer as well as winter, experienced in Antarctica and the 
neighbouring parts of the Southern Ocean, Greenland, and 
most of the Arctic Ocean, all those parts of the world, in fact, 
which are permanently covered with ice-sheets or closely 
packed floating ice. 

In studying the climates of past times the geography of 
which has been reconstructed, 1 it is useful to have some 
mathematical expression of this relationship between the 
temperature and the distribution of land and sea. Such an 
expression can be calculated in two different ways. We can 
either calculate the mean temperatures of each of a number 
of different parallels of latitude, and express the mean for any 
parallel in terms of its latitude <f> and the fraction n of the 
parallel which is covered by land, or alternatively, we can 
start with the mean temperatures of a number of individual 
points and represent them in terms of the latitude and the 
local land and sea distribution. The first method was initiated 
by J. D. Forbes (i) and further developed by R. Spitaler (2), 
who obtained an expression which, converted into Fahrenheit 
degrees, becomes : 

T (F.) =27 -6+32 cos^ + i3 cos 2^+35 n cos 2<. 

1 Sec Chapter XII. 


Since cos 2 <f> is positive between 45 N. and 45 S., negative 
from 45 to the poles, the term 35 n cos 2<f> indicates that the 
effect of land in low latitudes is to raise the temperature and 
in high latitudes to lower the temperature. This is simply a 
mathematical expression of the fact that an extensive land-mass 
tends to give a hot desert near the equator and a cold tundra 
near the poles. 

It is to be noted that Spitaler's formula is derived entirely 
from the present distribution of temperature and of land and 
sea, and therefore implicitly assumes the existence of the 
present system of winds, of ocean currents, and of ice. The 
fact that the formula applies reasonably well to both hemi- 
spheres, in spite of their very different configurations, shows 
that this is not necessarily a serious objection, but it is obvious 
that three conditions must be fulfilled before it can be applied 
to determinations of the temperature in other geological 
periods. First, there must be large areas of open ocean 
within the tropics, with free communication between low 
and middle latitudes, in order to allow for the great trans- 
ference of heat by ocean currents. Secondly, the land and 
sea distribution must not be of such a nature that the system 
of pressure and winds is radically different from the present 
system ; in the language of Chapter II., the planetary cir- 
culation must not be dominated by the geographical 
circulation. Thirdly, there must be extensive areas of 
floating ice. For example, Spitaler's formula gives for the 
Arctic Ocean a mean temperature of about 15 F., which is 
well below the freezing point of sea water. But we have seen 
in Chapter I. that if the " non-glacial " temperature of the 
polar regions could be raised by some 5 F. there would be no 
floating ice, and the mean annual temperature would be well 
above 32 F. The formula in its present form, therefore, 
does not apply to the " non-glacial " periods ; a better 
representation of the zonal distribution of temperature during 
these periods would be given by a formula of the type : 

T=T +# cos <f> b cos 2<f>-\-cn cos 2<, 

the negative sign of the term cos 2 < allowing for the decrease 
of the zonal contrast during these periods. 

For small changes of geography, however, Spitaler's 
formula gives a useful means of estimating the resulting 


changes of temperature. The result of a decrease of ten 
per cent, in the area of land in any latitude on the mean 
temperature of that latitude would be as follows : 

Latitude (degrees) o 10 20 30 40 50 60 70 80 
Change of tem- 
perature (F.) . -3-5 -3-3 -2-7 -1-7 -0-6 +0-6 +1-7 4-2-7 +3*3 

We will return to these figures later. 

We must now return to the second method of calculating 
geographical factors of temperature, in which the basis is 
the temperature distribution at a number of individual 
points instead of the mean temperature along whole parallels 
of latitude. This method has been extensively employed 
by F. Kerner and later by myself ; it can be used, not only for 
the whole world, but also for restricted areas, though as 
F. Kerner von Marilaun (Kerner- Marilaun) (3) points out 
with justice, the temperature of any given point depends 
not only on the geography of the region immediately 
surrounding the point, or even on the distribution of land and 
sea along the whole line of latitude ; we have to take into 
account conditions over the whole globe, as far as they affect 
the air and water circulation. He accordingly divides the 
geographical factors into local, or stenomorphogenous, and 
general or eurymorphogenous. Kerner has written a large 
number of palseoclimatological papers, of which I have 
selected three for discussion here. He subsequently put 
together his results in an important book (3). 

The first paper (4) deals with the winter climate of Europe 
during the Tertiary period. At present this temperature is 
governed on the west coast by the temperature of the Gulf 
Stream Drift, and decreases eastward in accordance with the 
distance from the Gulf Stream and the increasing continentality. 
He accordingly represents the temperature along any latitude 
by the equation : 

where t is the winter temperature of a European locality, 
T the winter temperature of the Gulf Stream Drift in the 
same latitude, d the distance of the point from the Gulf 
Stream Drift along that latitude (the " linear continentality "), 
L and / the percentages of land in large and small areas 
surrounding the place. A and B are constants, which were 


evaluated from the mean January temperatures reduced to 
sea-level taken at the intersections of every fifth degree of 
latitude and longitude from 35 to 55 N., and 20 W. to 
70 E. The best representations of the general and local 
continentalities L and / were determined empirically, and the 
definitions finally adopted were : for L the percentage of land 
in a twenty-degree " square " of latitude and longitude 
surrounding the point (/ 20 ), and for / the average of the 
percentages in five-degree and ten-degree squares (/ 5 and 
/ I0 ). A square in this sense is an area bounded by lines of 
latitude and longitude each covering the same number of 
degrees. The formula as finally calculated took the form 

^T-A( 4 +o- 2 AE)/ 20 -B- 5 (/ IO +/ 5 ), 

where AE is the longitude in degrees east of Greenwich. 
The values of A and B were calculated separately for each 
fifth degree of latitude. 

Fig. 14. Isotherm of 32 F. in Tertiary and Quaternary. 
After F. v. Kerner. 

By means of this formula and reconstructions of the land 
and sea distribution during six stages of the Tertiary period, 
the " stenomorphogenous " isotherms were reconstructed. 
The results are' summarised in a small figure showing the 
position of the January isotherm of 32 F. in each period, 
which is reproduced in Fig. 14. The results indicate a more 
favourable climate from the beginning of the Eocene up to 
and including the Miocene, a Pliocene climate differing little 
from the present and a less favourable climate of the early 
Quaternary in Western Europe. The differences from the 
present rarely exceed 10 F. in Central Europe, but east of 


the Caspian during the Middle Eocene and Oligocene the 
calculated January temperatures are nearly 30 F. above the 
present temperature. In the early Pleistocene the calculated 
temperatures are nowhere 5 F. below the present. The 
mean temperatures in January over the area covered by 
the figure, expressed as differences from the present mean 
temperature, are as follows : 

Latitude N 55 50 45 40 

F. F. F. F. 

Early Eocene . . . +2 + 6 f 10 +7 

Eocene +4 +9 +13 +-io 

Oligocene .... +5 +12 +13 f 9 

Miocene +r +3 +5 +3 

Pliocene 2 3 4 3 

Pleistocene . . . . i + i 4~ i - i 

Table 5. Kerner's calculated temperature differences, 

With regard to the distant or eurymorphogenous component, 
only a few qualitative remarks are possible. The submergence 
of the greater part of Florida would have increased the strength 
and velocity of the Gulf Stream somewhat, but the effect on 
the winter climate of Western Europe would have been slight. 
The existence of a land-bridge between Greenland and Europe, 
according to Semper, would raise the temperature off the 
west coast of France by some 13 F., but would lower the 
winter temperature of the Arctic regions. Semper thought 
that the Tertiary polar floras developed in a continental 
polar climate with hot summers, but this view is not tenable. 
For the warming effect of the " Indian Drift " which reached 
Central and Southern Europe from the Indian Ocean during 
the older Tertiary, Heer's simple assumption of a warming 
effect of 7 F. (5) is probably the best approximation that can 
be made at present. 

The winter temperatures deduced by Heer from the fossil 
plants in the Miocene are from 5-io F. higher than the 
temperatures due to the local geography during that period, 
but this difference is sufficiently accounted for by the effects 
of more distant changes, and particularly by the warming 
effect of the Indian Drift. Kerner does not consider the 
possible effect of a change from " glacial " to " non-glacial " 


The second paper by F. Kerner (6) extends this study 
to the Arctic regions. The distribution of the January 
temperatures along each parallel of latitude in the Arctic 
Ocean is expressed in the form 


where w is the warming effect of a gap ten degrees of longitude 
broad, open to the world oceans, and k is the cooling effect 
of a ten-degree barrier between the Arctic and the open ocean 
farther south. The values of the terms w and k are deduced 
from the distribution of temperature in January under present 
geographical conditions, and from these values and the land 
and sea distribution reconstructed for the Middle Eocene the 
isotherms for that period are reconstructed. Some attempt 
is made to allow also for a decreased cooling of the Gulf 
Stream by the cold Labrador Current, but the calculations 
implicitly assume a large area of floating ice in the Arctic 
Ocean. The reconstructed temperatures are still below 
freezing point over most of the region north of 70 N., but they 
are well above the present temperatures, and it seems probable 
that the improvement was sufficiently great to raise the mean 
" non-glacial " temperature at the pole above the freezing 
point. In accordance with the argument of Chapter I., 
this would mean that there would be no floating ice-cap, 
and a totally different climatic regime would come into force. 
This question can be better investigated on the basis of 
some more recent work of Kerner' s (7) in which he returns 
to the discussion of Arctic temperatures. The first part of 
this paper, dealing with the akryogenous or " non-glacial " 
marine climate of an open polar ocean, has already been 
referred to in Chapter I. In the second part, Kerner analyses 
the distribution of January temperature at present along the 
75th parallel of latitude. This parallel runs mainly over the 
ocean, the only land which it crosses being Novaja Zemlya, 
the Taimyr Peninsula, the New Siberian Islands, parts of the 
Arctic Archipelago, and Greenland, but over the greater part 
of its course it has extensive land-masses Europe, Asia, and 
North America within five degrees to the southward. There 
are only two gaps in this surrounding land-ring, the very 
narrow and shallow Bering Strait, through which very little 
warm sea water penetrates at any season of the year and none 


at all in winter, and the broad Atlantic gap through which 
the Gulf Stream Drift makes its way. The surface of the 
Arctic Ocean is cooled in winter by the cold winds from the 
winter anticyclones over the great continents, and warmed 
by the Gulf Stream Drift, and the January temperature at 
any point along the 75th parallel therefore depends on the 
amount of land to the southward and the distance from the 
Atlantic gap through which the warm water enters. These 
two factors are named by Kerner the " Continental " term 
K, and the " Separation " term S. 

To get the matter clear, imagine a one-roomed cottage 
with thin walls, against the outside of which snow is banked, 
making them very cold, while in one wall there is a fire. 
The temperature at any point near one of the walls is then 
determined by the distance from the wall and the distance 
which separates the point from the fire. The walls represent 
the cold continents and the fire the Gulf Stream Drift. Kerner 
finds that the January temperature of a point on the 75th 
parallel is given (in Fahrenheit degrees) by the expression : 

T=45-i-o7 S 1-9 K. 

The precise evaluation of the " Separation " term S and 
the " Continental " term K is somewhat complex and need 
not be gone into here. 

According to this formula, the most favourable distribution 
of land conceivable for high polar temperatures would be a 
number of long, narrow islands extending from low to high 
latitudes, separated by wide, deep channels. This distribution 
may have actually occurred at some stage in the middle of the 
Palaeozoic period, and it was approached to some extent during 
parts of the Mesozoic and Tertiary periods. Two examples 
are considered by Kerner, the Middle Eocene and the Upper 
Jurassic. In the Middle Eocene, according to a recon- 
struction by Matthew, the circum-polar ring was broken by 
three broad gaps, an enlarged Bering Strait, the present 
Atlantic gap, and the Obic Sea which separated Europe from 
Asia. In the Upper Jurassic, according to a reconstruction 
by Uhlig, the Atlantic gap was replaced by the " Shetland 
Strait " farther west, and there was in addition a fourth gap, 
the Jana Sea, between 120 and 140 E. 


From his formula, Kerner calculates the mean January 
temperatures for the 75th parallel to have been as follows : 

Present. Mid-Eocene. Upper Jurassic. 

F. F. F. 

-20-7 +7-8 +18-5 

All the figures for individual longitudes are higher than 
the present ones in the same longitudes, but none of them 
exceed 32 F., and they still represent a very severe climate. 
We saw in the Introduction that during both these periods, 
vegetation of a sub-tropical or warm-temperate aspect 
flourished at several points north of 70 N., and Kerner 
considers that either the plant evidence is not reliable or the 
solar heat must have been greater. 

The real reason for the discrepancy is that, in spite of 
his discussion of akryogenous (" non-glacial ") temperatures 
in the first part of the paper, Kerner employs the actual 
distribution of temperature as the basis of his calculations, 
instead of the " non-glacial " distribution. Thus his calcu- 
lated figures imply the presence of a large amount of floating 
ice and ice-cold water in the Arctic Ocean, whereas we have 
seen that, given a " non-glacial " temperature only five 
degrees above the present, there would be no ice, and the 
water cooled by radiation would at once sink to the bottom. 
Evidently we have to recalculate his figures on a " non-glacial " 

Kerner's Middle Eocene mean January temperature is 
28-5 F. above the present mean in 75 N. Of this increase 
we find that 6-9 F. is accounted for by the decrease in 
continentality and the remaining 2i-6F. by the increased 
influx of warm ocean currents (Separation effect). The small 
continentality effect may be allowed to stand, but the Separa- 
tion effect requires some modification. Suppose a warm current 
at temperature t introduced into a mass of water at tempera- 
ture ?, the resulting mean temperature of the water-mass 
being T. Then we can suppose that the warming effect (T t 1 ) 
is proportional to (tt 1 ), and we can write T^-f+c^ t) 
where c is a constant fraction, depending on the volumes 
of the mass of cold water and of the warm current. Hence 
the present Gulf Stream Drift would have a smaller warming 
effect on a non-glacial Arctic Ocean than on the present 


glacial Arctic Ocean. On the other hand, if there were no 
Arctic ice there would be no cold East Greenland and 
Labrador Currents. From a study of the sea surface isotherms 
of the North Atlantic, we find that in January the temperature 
along the centre of the Gulf Stream is 71 F. in latitude 
30 N. and 64 F. in 38 N., a fall of 0-9 F. per degree. 
From 38 to 43 N. on the other hand temperature falls by 
about 22 F. in only 5 degrees. Of this fall, only about 5 F. 
can be due to the normal fall with latitude, and the remaining 
17 F. is due to admixture with the cold water of the Labrador 
Current. That is, the present January sea surface isotherm of 
about 32 F. in 75 N., 10 E., would be replaced by one of 
49 F. Now we have the following data for a recalculation 
of the warming effect of the Gulf Stream Drift in a non-glacial 
polar basin with the present configuration : 

" Glacial " " Non-glacial " 

Conditions. Conditions. 

F. F. 

Temperature of Gulf Stream Drift, / 32 49 

Temperature of Arctic Ocean, T . . - 18 25 

Difference (/ T) 50 24 

Table 6. " Glacial " and " Non-glacial " temperatures. 

The total difference (tt 1 ) is proportional to (/- T), so 
that we have to multiply the coefficient i 07 of Kerner's 
"Separation" effect S by 24/50 or approximately 0-5 in 
order to correct this factor for a non-glacial ocean with the 
present land and sea distribution. If, now, we introduce a 
second current equivalent to the Gulf Stream Drift, the 
additional warming effect will be proportional, not to (t t 1 } 
but to (/ T), and the resulting temperature T 1 will be equal 
to ^+2^(/ / x ) c*(t t 1 }, i.e., the increase is something less 
than twice that due to a single Gulf Stream Drift. The 
constant c is small, so that the additional term is not im- 
portant ; moreover, Kerner's method of calculation makes 
some allowance for it, but in order to be on the safe side I 
have reduced the factor 0-5 to 0-4. The increase in the 
non-glacial temperature of the Middle Eocene, due to the 
change in the value of S, should therefore be only 8 '6 F., 
instead of 21 -6 F., making, with the increase of 6*9 F. due 
to the decreased continentality, the total increase in the 


non-glacial January temperature during this period 15*5 F. 
This has to be added to the mean January " non-glacial " 
temperature at present, 25 F., raising the Middle Eocene 
temperature of the Arctic Ocean in 75 N. to 40-5 F. This 
is well above the freezing point of sea water, so that there is 
no ice, and 40-5 F. is also the real January temperature. 

For the Upper Jurassic, Kerner obtains a January tem- 
perature in latitude 75 N. of 39-2 F. above the present. 
Of this amount, his calculations show that 6-3 F. is due 
to the change in the Continental effect and 32 9 F. to the 
change in the Separation effect. Applying the correction 
for a non-glacial ocean by multiplying by 0-4, the latter 
quantity becomes 13*2 F., making with the change in the 
continental effect a total increase of the " non-glacial " tem- 
perature of 19-5 F. Added to the present non-glacial 
temperature of 25 F., this makes the Upper Jurassic January 
temperature 44-5 F. in latitude 75 N., a figure which is 
very near the present mean January temperature of South- 
west England, so that these temperatures are quite consistent 
with the remains of the vegetation discovered by geologists. 
While I make no pretence as to the absolute accuracy of these 
computed figures, I do claim that they are so far above the 
freezing point of sea water that the strictest revision is unlikely 
to reduce them below it, and that the case for an ice-free 
Arctic Ocean during some periods of geological time is 
thereby established. 

It has been suggested to me that the figure of 17 F. 
adopted for the cooling of the present Gulf Stream Drift by 
ice and ice-cooled Arctic water is too great, and this will 
serve as an example of the effect produced on the calculation 
by a modification of the basis. Let us reduce this figure 
from 17 F. to 9 F., which is almost certainly too small. 
Repeating the calculations on this new basis, we find that the 
mean January temperature in 75 N. comes out as 37-5 F. 
during the Middle Eocene and 40 F. during the Upper 
Jurassic. These figures are well above the freezing point, 
so that the Arctic Ocean would still be non-glacial. 

Another objection which may be raised to the way in 
which these high polar temperatures have been obtained 
is that they depend on the existence of powerful ocean currents, 
which in turn depend on the planetary wind system, and that 


a warming up of the Arctic Ocean without a corresponding 
change in the temperature of the equatorial zone would 
result in a decreased strength of the wind-driven ocean 
currents. This possible objection has already been dealt 
with in the chapter on Pressure and Winds, page 53, where 
it was shown that there is a critical point in the planetary 
circulation. With a temperature difference above this critical 
value, the winds are largely directed outwards from the pole, 
ocean currents have difficulty in penetrating into the Arctic 
basin (barely 2 per cent, of the water in the Gulf Stream off 
Florida reaches the Arctic Ocean), and their temperature 
is also greatly lowered by the cold winds. With a temperature 
difference below the critical value, the winds are directed 
inwards towards the pole, and the volume and temperature 
of the ocean currents are maintained with much smaller loss. 
Hence the oceanic circulation induced by high polar tem- 
peratures would assist in maintaining those temperatures ; 
similarly, the oceanic circulation with low polar temperatures 
would help to keep the temperature low. 

The interesting question arises, has the temperature of 
the Arctic Ocean risen above the critical point at any stage 
of post-glacial time ? I think there is no doubt that it has (8). 
During the " Climatic Optimum " there was a rich flora in 
Spitsbergen, while the fossil marine mollusca indicate a coastal 
sea temperature much higher than the present in all the 
Arctic lands which are at present dominated by sea ice, 
including Iceland and Greenland. The " Climatic Opti- 
mum " was experienced also over most of Europe, where 
it has been studied more closely than in the Arctic lands. 
In Scandinavia the warm period appears to have begun 
rather suddenly during a period of increased vigour in the 
circulation of the Atlantic Ocean (" Atlantic " or " Maritime 
Phase"). During this period, owing to submergence, the 
Baltic lay more open to the Atlantic than at present, and a 
maritime climate extended as far as the coast of Finland. 
Depressions passed readily across Denmark and along the 
German coast, so that in Northern Europe the Atlantic period 
had a greater rainfall than the present, with milder winters 
and cooler summers. About 3000 B.C. the connexion between 
the Baltic and the Atlantic again became restricted, and there 
was a slight extension of the area of the British Isles. These 


changes were associated with a notable difference of climate. 
Depressions seem to have favoured a northerly track into 
the Arctic Ocean, and the British Isles, Western and Central 
Europe became much more continental, with very warm 
summers ; the winters do not seem to have been any colder 
than at present. These conditions were very marked in 
Switzerland, where settlements occurred at very high levels 
and there was much traffic over passes which are now occupied 
by glaciers. In Spitsbergen the " ice floor " melted com- 
pletely. The mean annual temperature at Green Harbour 
is at present 19 F., so that there must have been a rise of 
mean temperature by at least 13 F. in this part of the Arctic. 
The favourable climate lasted until about 3000 B.C., deterior- 
ated slowly until about 500 B.C. and then came to an abrupt 
end. The change of climate for the worse was very rapid, 
and, according to H. Gams and R. Nordhagen (9), in the 
Alps it " had the appearance of a catastrophe." 

My reading of this history is that the increased circulation 
of the Atlantic period swept away the ice from the Arctic 
Ocean (though apparently not from the channels among 
the islands north of America and Greenland). After the 
passing of the Atlantic period, the Arctic Ocean became 
somewhat cooler, but, being still free of ice, was very stormy, 
and this storminess itself maintained the winter temperature 
above the critical point and prevented the ice-cap from 
beginning to form. Then came an unusually quiet cold 
winter, the ice-cap obtained a footing, and perhaps in the 
course of a single season covered the greater part of the Arctic 
Ocean. The result was a sudden great change in the climate 
of Europe ; the conditions of to-day came in " with the 
appearance of a catastrophe." The ice-cap, once formed, 
kept the winter temperature below the critical point by its 
own power of persistence. 

It is possible that the Arctic Ocean again became free of 
ice during historic times, from about the fifth to the tenth 
or eleventh centuries of the Christian era. O. Pettersson (10) 
makes out a good case for the absence of sea ice in the East 
Greenland Current during the latter part of this period. 
His map of the old Norse sailing routes shows a track direct 
from Iceland to the east coast of Greenland in latitude 66 N., 
then down the coast to Cape Farewell, and up the west coast. 


According to the documentary evidence which he adduces, 
this route was followed until nearly A.D. 1200, and for most 
of the period there is no mention of ice in any of the numerous 
descriptions. From Greenland the Norsemen sailed to 
Wineland (on the coast of North America), and again there 
is no mention of ice. Recently this question has been re- 
investigated by L. Koch (n). From an exhaustive study of 
historical records of the ice off East Greenland and Iceland he 
concludes definitely that from A.D. 800 to 1200 there was 
scarcely any summer ice near Iceland. This is very striking ; 
Pettersson's own inference is that the ice did not then come 
so far south as it does now, and it seems probable that the 
Arctic Ocean was, if not ice-free, at least in the intermediate 
or " semi-glacial " condition described in Chapter I., in 
which a small cap formed in winter but disappeared completely 
in summer. This question is discussed further in Part III. 

In 1917 I made my first incursion into the subject of 
continentality and temperature (12). This paper dealt with 
the region between 40 and 60 N. and between the Atlantic 
coast of Europe and 90 E., that is, practically the same 
region as the first of Kerner's papers (4), which I had not 
then seen. Fifty-six stations were selected in this area, and for 
these were found the height above sea-level, the mean tem- 
peratures of January and July, the " continentality " and the 
radiation receive . The " continentality " was measured 
as the percentage of land in a five-degree circle, in a zone 
between five- and ten-degree circles, and in a zone between 
ten- and twenty-degree circles surrounding each station, 
these measures being termed respectively G 5 , C 5 . IO , and C I0 . 20 . 
It was also found necessary to introduce a Gulf Stream com- 
ponent into the January temperatures north of 50 N., the 
temperature decreasing by 0-6 G. for every hundred kilo- 
metres, or 1-7 F. for every hundred miles, east of a great 
circle through Valentia in South-west Ireland, and touching 
the north-west coast of Norway. 

The most important result of the investigation was to 
show that the January temperature of Europe is much more 
closely related to the land and sea distribution than to the 
amount of heat received from the sun. The solar heat is 
the same at all points on the same line of latitude (apart from 
the effect of cloudiness, which was not discussed but which 


itself depends on the land and sea distribution) and decreases 
rapidly from south to north, so that if the temperatures were 
governed by this cause only, the isotherms should run east 
and west. It is found that actually they run nearly north 
and south, the rate of temperature decrease eastwards from 
the coast towards Russia being much more rapid than the rate 
of decrease northwards from Southern to Northern Europe. 
In July the effect of land and sea distribution is less marked 



Radius of Inland in Degrees 
5 10 16 6 




500 1000 1500 2000 

Ares of Island In Tnousands 
of Square Miles 

Fig. 15. Change of temperature due to formation 
of an island. 

and is slightly exceeded by the effect of solar radiation, so 
that the isotherms run more east-west than north-south. 

Another interesting result concerns the effect of land- 
masses of different areas. Suppose a circular island were 
to form in mid-ocean in about latitude 60 N., and to increase 
gradually in size until it reached a radius of twenty degrees 
of arc or an area of about two million square miles. The 
resulting changes of temperature at a point on the edge are 
shown in Fig. 15. With increasing area the January tempera- 
ture (lower curve) would fall and the July temperature (upper 
curve) would rise. From the figures obtained it appears thai 



the fall of temperature in January would be very slow at first, 
being only 0-5 C. or 0-9 F. by the time the island had 
reached a radius of five degrees of arc (area about 375,000 
square miles). As the island increased in size the fall of tem- 
perature would then become very rapid, until by the time the 
radius was ten degrees (area 1,500,000 square miles) it would 
amount to 22 C. (40 F.). After this the cooling would again 
increase more slowly with increasing area. 

This curious curve is connected with the influence of the 
island on the atmospheric circulation. In Chapter II. we 
found that the effect of a small ice-covered island on the 
atmospheric circulation is slight the storms sweep across 
it with very little hindrance. A larger island modifies the 
distribution of pressure, and an island with a radius of about 
450 to 500 miles (about seven degrees) begins to develop a 
winter anticyclone. With a radius of ten degrees this winter 
anticyclone becomes semi-permanent and dominates the 
pressure distribution, and any further increase of radius 
merely results in a slow increase of intensity. While the radius 
is not more than five degrees (350 miles), the general meteoro- 
logical conditions are unaltered and the cooling effect of land 
is due to its lower specific heat, which causes it to lose its 
summer heat more rapidly than does water, but with a radius 
of seven degrees (480 miles) or more, the winter anticyclone 
prevents the influx of heat from the neighbouring sea, and 
cooling by radiation proceeds rapidly. 

Summer conditions are very different. As the island 
increases in size, the July temperature rises rapidly at first, 
until the radius has reached about five degrees (350 miles). 
Up to this point the sole effect is probably the absorption 
of the sun's heat by the soil and its transference to the lower 
layers of air by conduction. When the radius reaches seven 
degrees (480 miles) the increase of temperature becomes 
slower, and may even cease altogether ; this is probably due 
to a sea-breeze effect which lowers the afternoon temperature 
considerably. After the radius has exceeded about twelve 
degrees (830 miles) the July temperature begins to rise again 
more rapidly, but with a radius of twenty degrees the warming 
effect at the centre is only about n G. (20 F.). 

The broken line in Fig. 15 shows the effect of the island 
on the mean annual temperature (mean of January and 


July). The annual temperature is raised slightly by the 
introduction of an island with a radius of less than seven 
degrees (480 miles), the maximum effect, a rise of rather 
more than i C. (2 F.) in the mean temperature, occurring 
in the centre of an island of radius five degrees (350 miles). 
With greater areas the mean annual temperature is lowered. 
The distribution of land most favourable for high average 
temperatures is therefore a number of small islands each 
about 700 miles across, a condition which was frequently 

Fig. 1 6. Observed, and calculated temperature 
changes, Litorina period, January. 

approached in Europe during the early part of the Tertiary 

This investigation concluded with an attempt to recon- 
struct from the changes in the land and sea distribution 
during the Litorina post-glacial submergence of Scandinavia 
(which occurred during the Atlantic period), the changes 
in the mean temperatures of January and July. For this 
purpose the present mean temperatures of those months at 
a large number of Scandinavian and Baltic stations were 
compared with the percentage of land in a five-degree circle, 


and it was found that a decrease of one per cent, in the 
continentality raised the January temperature by 0*20 G. 
(0-36 F.) and lowered that of July by 0-06 G. (0-11 F.). 
On this basis the change of temperature over the Northern 
Baltic was calculated as +3 C. (+5 F.) in January and 
i C. ( 2 F.) in July. The calculated temperatures in 
different districts are compared with those deduced by various 
authors from the fauna and flora in Table 7 ; the results for 
January are shown graphically in Fig. 16, in which the 
calculated changes are shown by lines of equal temperature 
change, and the variations from present conditions required by 
geologists are shown by the figures. A variation of uncertain 
amount is indicated by the sign + or without a figure. 
The agreement is good on the whole. The actual type of 
change in the direction of a more insular climate, warmer on 
the whole, is in perfect agreement. Many of the botanists 
comment on the prolongation of the autumn into the present 
winter, which is especially characteristic of the change to a 
more insular climate. The amounts of the change are also 
in good agreement except in the Christiania region and in 
North Denmark, where the geologists require a greater change 
than would be inferred from the change of continentality. 
This is probably accounted for by the greater freedom of 
ingress which the more open seas allowed to the warm waters 
of the Gulf Drift, but this point is discussed in detail later. 
In 1918 I was able to extend the study of " Continentality 
and Temperature " to embrace the greater part of the 
world (13). The method adopted was extremely simple, 
perhaps rather too simple. From various sources I obtained 
the average mean temperature in January and July at each 
point of intersection of the ten-degree co-ordinates of latitude 
and longitude over both land and sea. On a globe 1 a ten- 
degree circle was drawn round each of these points, divided 
into east and west semi-circles, and the amount of land in 
each semicircle was measured and expressed as a percentage 
of the area of the semicircle. The area of ice in the whole 
circle was also measured and expressed as a percentage of the 
area of the whole circle ; this area included both land ice and 

sea ice. In winter the area of sea ice is many times the area 


1 The octagonal globe employed in the Meteorological Office for work con- 
nected with the Roseau Mondial was utilised for this purpose. 




Inferences from Fauna 
and Flora. 





Rekstadt and 

No trace of warm period. 

+ 1-0 




(65 N.). 


J. Rekstadt. 

Climate " not greatly dif- 

-f I 'O 


ferent from present." 

Bergen . 

C. F. Kolderup. 

Climate " somewhat mil- 

+ 0-2 


der than present." 


K. Bjorlykke. 

No marked warm period. 



(S.W. Nor- 


Sorland (S. 

D. Danielsen. 

Climate similar to present. 



Norway) . 

J. Holmboe. 

" Somewhat warmer than 


Christian ia 

C. Brogger. 

Rise of 2 in mean annual 





J. Holmboe. 

Rise of 1-9 to 2-2 in 

mean annual tempera- 

ture. Climate more 



V. Nordmann. 

A damp, warm period 

+ 0-3 



(warmer winters, sum- 

mers unchanged). 


G. Andersson. 

About 2 warmer. More 

-f i to 




+ 3 

to - i 

R. Sernander. 

Insular climate. 

L. von Post. 

Warm, moist. 

North Germany. 

Several authors. 

The difference, if any, was 

o-o to 

o-o to 

in the direction of a more 


-f 0-2 

continental climate. 

East Baltic. 


Damp, warm (climate of 



West European coasts). 

H. Lindberg. 

Finland had a more in- 


- I -0 

sular climate. 

Table 7. Comparison of actual and calculated 
changes of climate. 


of land ice, and the results are taken as applying to the former. 
The effects of each of these variables land to the west, 
land to the east, and ice, independently of the other two 
were then worked out by the method of correlation. The 
results can be set out in general terms as follows : 

1. In winter, the effect of land to the west is always to 
lower temperature. This holds in every latitude except 
10 S. and 20 S. 

2. In winter, the effect of land to the east is almost 
negligible. The only important exception to this 
rule is in 70 N. latitude, which may be considered 
as coming within a belt of polar east winds. 

3. In summer, the general effect of land whether to the 
east or west is to raise temperature, but the effect is 
nowhere anything like so marked as the opposite effect 
of land to the west in winter. 

4. The effect of ice, in the few cases in which it is possible 
to measure it, is invariably to lower temperature. 

5. The temperature even of a point in mid-ocean in any 
latitude is modified by the presence of land along other 
parts of that parallel of latitude. The January tem- 
perature of a point in mid-Atlantic in latitude 60 N., 
for instance, is higher at present than it would be if 
the North Pacific were occupied by land. 

These general conclusions could have been arrived at 
without a laborious statistical analysis, but the latter was 
necessary to reduce them to figures, and so make possible 
calculations of the thermal effect of changes in the land and 
sea distribution. This calculation is carried out by means 
of the formula : 

where T is the temperature of the point required, Z is the 
" zonal temperature " (see below), L w is the percentage of 
land in the semicircle to the west, L E the percentage of land 
in the semicircle to the east, and I the percentage of ice in 
the whole circle, a, i, and c are constants for any particular 


The " zonal temperature " Z is not a fixed quantity for 
any particular latitude ; it depends on the amount of land 
in the neighbourhood of that latitude, a zone which is mainly 
land-covered having a lower zonal temperature than a zone 
which is mainly occupied by water, but the relationship is 
not simple. 1 

The coefficients a, b, and c were obtained by purely statistical 
methods. They are as follows : 


b c 



Land to 

Land to 

Land to 

Land to 



east. Ice. 



east. Ice. 

70 N. 


0-20 0-46 

70 N. 


4-0-02 -0-16 



-o-oi -0-07 






4-0-09 0-09 

5 I 























10 N. 










! 10 s. 





















Table 8. Effect of one per cent, of land to the west, of land to the 
east, and of ice on temperature. 

It should be noticed that in this table the unit area of 
ice, one per cent, of the whole circle, is twice that of the 
unit area of land, one per cent, of a semicircle. The two 
figures can be made comparable by adding together the 
two land coefficients, thus : 

January. July- 

70 N. 60 N. 50 N. 70 N. 

Effect of land . . 0*63 0-32 0*20 40*04 

Effect of ice . . . 0*46 0*07 0*09 0*16 

The cooling effect of ice in winter is apparently less than 
that of land. This is because most of the area shown as 
occupied by ice is sea which is covered by more or less 

Winter : 

Z - 70(0-95-003 0) log L- 

tan* L 

Summer : 


= 30(1 -05- cos 0) log L4- 

tan 2 




cattcred drift ice and icebergs, that is, the surface is partly 
ice and partly water. In high latitudes in winter the land is 
usually snow covered, and it is easy to see that this snow 
surface must have a greater cooling effect than scattered 
sea ice. For ice-sheets over land it would be best to adopt 
the coefficients of land in winter. 

These formulae, being based on the present land and sea 
distribution, could not be employed in calculating the 
distribution of temperature during periods with a radically 
different distribution of the continents, but it seems legitimate 
to employ them to calculate the differences from the present 

Fig. 17. Lines of equal change of land level in Quaternary, 

and changes of land and sea distribution. Additional land 

shaded, additional sea black. 

of the temperature distribution in a geological period during 
which the main outlines of the geography were the same as 
at present. In doing this it is best to measure the differences 
in the land and sea distribution from the present, and to 
calculate from these the differences in the temperature dis- 
tribution, rather than to attempt to work ab initio. By working 
only with differences, we preserve intact the local peculiarities 
of climate and minimise the risk of error. The formulae were 
applied in this way to a reconstruction of the land and sea 
distribution during the early part of the Quaternary period. 
This was the culmination of a great period of elevation in 
cold temperate regions, with corresponding depression in 


the tropics. I have represented the differences in level and 
land and sea distribution between that time and the present 
in Fig. 17. The lines of equal change of height were drawn 
from plotted figures accumulated from a great variety of 
sources. The restoration of the land and sea distribution is 
based mainly on the change of height used in conjunction 
with relief and bathymetrical maps, but in a few cases the 
actual ancient shore line has been traced ; it depends on the 
assumption that the continents held their present positions. 
The hypothetical restoration of Antarctica is based on well- 
known bio-geographical data, much of which has been admir- 
ably summarised by C. Hedley (14). Bio-geographical data 
have also been used as additional criteria in a few cases, such 
as the separation of Madagascar from Africa and New Zealand 
from Australia, or the union of Siberia to Alaska and of Japan 
to the mainland. It is also necessary to remark that there is 
not always evidence that the changes were strictly contem- 
poraneous, but there is enough to show that the map is 
sufficiently correct to form the basis for a discussion. From 
this map and the formulae it is evident that, even without 
an increase in the glaciation, the fall of temperature in winter 
outside the latitudes of 40 must have been very considerable* 
This fall was still further augmented by the great increase in 
the altitude of these regions. Now we know that, except in 
parts of Asia, practically the whole land surface north of 50 N. 
was glaciated, so that as a rough approximation we may 
assume that glaciers formed wherever the mean annual tem- 
perature fell below 32 F. (the problem of precipitation is 
dealt with in Chapter IX.). At first the increase of the land 
area would have raised the summer temperatures, but as the 
snow-cover began to persist through the year and form per- 
manent ice-sheets, this excess disappeared and was replaced by 
a deficit. Taking the coefficient of land ice in summer as 
0-16 C. for one per cent, of a ten-degree circle, and the 
average coefficient of land as +0-05 C., every increase of four 
per cent, in the portion of a ten-degree circle of land covered 
by ice lowered its temperature by 0-8 C. (1-4 F.), and where 
the ice extended on to the sea the lowering was 0*6 C. 
(i-iF.); consequently, within the borders of the great 
ice-sheets the lowering of temperature in July amounted to 
nearly 20 C., an amount sufficient to enable the accumulation 


of ice to continue in summer as well as in winter. In January 
the change probably made little difference. 

In Figs. 1 8 and 19 are shown the calculated differences 
of temperature from the present, both before the formations 
of the ice-sheets and at their maximum extension. The lines 
are lines of equal difference of temperature from present 
conditions in the same months. The calculated differences 
at a few points in the Northern Hemisphere are set out in 

Fig. 1 8. Changes of temperature due to geographical 
changes, January. 

Table 9, with, for comparison, the differences at the same 
points deduced from the geological and biological evidence : 



Inferred Fall. 

Calculated Fall. 
Jan. July. Mean. 



J. Geikie. More than 20 36 


East Anglia. 

C. Reid. 





Penck and 










Table 9. Comparison of calculated change of temperature 
with that inferred from geological evidence. 

The agreement is quite good, and seems to show that the 
decrease of temperature during the Quaternary Ice-Age was 
completely accounted for by the changes in the distribution 
of land and sea and the effect of the ice itself. This does not 


J 55 

necessarily mean, however, that an increase in the land 
area in high latitudes would by itself suffice to initiate a 
glaciation ; the winter temperatures would be lower than 
at present, but the summer temperatures would be higher, 
and on low ground the winter snowfall would not 
survive the hot summer that is why Siberia is not glaciated. 
Probably, glaciers can only originate on an area of high ground, 
but the presence of a large continental region with a low winter 
temperature is essential if a local glaciation of the mountain- 
valley type is to develop into a regional glaciation of the ice- 
sheet type. This is clearly illustrated in the glacial history of 

Fig. 19. Changes of temperature due to geographical 
changes, July. 

Europe ; the Quaternary glaciation began in the mountains 
of Norway, but the extension of the ice half-way across 
Europe was possible only because, owing to elevation, there 
was a large area of high continentality to the eastward. The 
beginning of glaciation in Norway was due to increased 
elevation bringing a larger area above the snow-line, and 
perhaps also to the shutting out of the Gulf Stream Drift. 
Once the glaciers had reached a certain size they became 
independent of the elevation of the ground in their centres, 
and their extension was governed, first, by the amount of 
snowfall available for their nourishment, and secondly, the 
snowfall being sufficient, by the balance between the cooling- 
power of the ice and the natural or " non-glacial " temperature 
of the regions into which they intruded. 


It will be interesting to compare the results of a ten per 
cent, decrease in the amount of land in each zone of latitude, 
calculated from my formula, with those obtained by means 
of Spitaler's formula given on page 132. According to my 
formula, the change of temperature falls into two parts, a 
local part due to the local changes in the land and sea 
distribution, and a general part due to the change in the 
general zonal temperature. For example, if owing to local 
elevation the Faroe group became a single large island of the 
size of Iceland, the result would be two-fold a large local 
decrease in the winter temperature of the site of the new island 
and of the surrounding area, and a small general decrease 
in the temperature of the whole zone between 50 and 70 
North latitude. 

The effect of a change from 50 to 40 in the percentage of 
land in any zone of latitude on the mean temperature of the 
whole zone would be as follows : 

Latitude .... o 10 20 30 40* 50 60 70 

January, F. . . . 0-5 0-4 -fo-i -f i i +2-6 4-4-7 +7*3 (+16) 

July, F 0-3 0-4 i -a 1-3 2-2 2-4 3-3 3'4 

Mean, F 0-4 0-4 0-5 o-i 4-0-2 4-i*i +2*0 ( + 6) 

Spitaler's Mean, F. 3-5 3-3 2-7 1-7 0*6 -fo-6 4-1*7 4-2-7 

Table 10. Effect of a decrease of 10 per cent, in the amount 
of land in a zone of latitude. 

The figures for 70 calculated by my formula are uncertain, 
but in 60 my result is in good agreement with Spitaler's, 
which was obtained by quite different methods. The chief 
difference in the results occurs in low latitudes, where the 
effect of land area is much less important according to 
my results than according to Spitaler's. 

The essential point for the theory of climatic variations 
is that, according to either computation, the warming effect 
of a decrease in the land area becomes greater as we get 
nearer to the poles. In summer, a decrease in the area of 
(unglaciated) land results in a slight decrease of temperature, 
and the net increase during the year is entirely due to the very 
large increase in winter. If the percentage land covering 
north of 60 N. decreased by ten, the result would be a rise 
in the " non-glacial " January temperature by at least 7 F., 
which would be sufficient to bring it above the freezing point 
and so introduce a " non-glacial " climate. This is the 


general case, of which the two examples Middle Eocene and 
Upper Jurassic discussed by F. Kerner are special instances. 
The conclusion to which we are brought, therefore, is that 
moderate changes in the land and sea distribution, such as 
have occurred frequently enough in geological times, are 
amply sufficient to bridge the gap between non-glacial and 
glacial climates, or between warm and cold geological periods, 
and that extraneous aids, such as variations of solar radiation 
or changes in the astronomical climate, while possible causes, 
are not necessary conditions. 


(1) FORBES, J. D. " Inquiries about terrestrial temperature." Edinburgh, 

Trans. R. Soc. y 22, 1861, p. 75. 

(2) SPITALER, R. " Die Warmevertheilung auf dcr Erdobcrflachc." Wien, 

Denkschr. K. Akad., 51, 1886, Abt. 2, p. i. 

(3) KERNER- MARILAUN, F. " Palaoklimatologie." Berlin (Gebr. Born- 

traeger), 1930. 

(4) KERNER, F. " Synthese der morphogenen Winterklimate Europas zur 

Tertiarzeit." Wien, 1913. 

(5) HEER, O. " Untersuchungen iiber das Klima und die Vcgetationsverhalt- 

nisse des Tertiarlandes." Winterthur, 1860. 

(6) KERNER, F. " Klimatogenetische Betrachtungen zu W. W. Matthews, 

' Hypothetical outlines of the continents in Tertiary times V Wien, 
Verh. k. k. geol. Reiclisanstalt, 1910, p. 259. 

(7) KERNER, F. " Das akryogene Seeklima und seine Bedcutung fur 

geologischen Probleme der Arktis." Wien, Sitzungsber. Ak. Wiss., 131, 
1922, p. 153. 

(8) BROOKS, C. E. P. " The problem of warm polar climates." London, 

Q,. J. R. Meteor. Soc., 51, 1925, p. 83. 

(9) GAMS, H., and R. NORDHAGEN. " Postglaziale Klimaanderungen und 

Erdkrustenbewegungen in Mitteleuropa." Miinchen, Geogr. Gesellsch. 

Landesk. Forschungen, H. 25, 1923. 
(10) PETTERSSON, O. " Climatic variations in historic and prehistoric time." 

Svenska Hydrogr.-Biol. Komm. Skriften, 5, 1914. 
(n) KOCH, L. "The East Greenland ice." Copenhagen, Komm. Viden- 

skabelige Under so gelser i Grdnland, Kobenhavn, 1945. 

(12) BROOKS, C. E. P. " Continentality and temperature." London, 

Q,. J. R. Meteor. Soc., 43, 1917, p. 169. 

(13) BROOKS, C. E. P. " Continentality and temperature." (Second paper). 

" The effect of latitude on the influence of Continentality on temperature." 
London, Q. J. R. Meteor. Soc., 44, 1918, p. 263. 

(14) HEDLEY, C. " The palaeographical relations of Antarctica." London, 
Proc. Linnaan Soc., 124, 1911-12, p. 80. 


THE various forms in which water falls from the sky, 
of which the most frequent are rain, snow, and hail, 
are conveniently grouped together under the term 
precipitation. The precipitation is occasionally slightly aug- 
mented by other forms, dew and hoar-frost, and on the arid 
western coast of South America there are mountain plants 
which live on the moisture they derive from mist, but, 
practically speaking, the three forms first mentioned are 
the only ones which need be considered. Precipitation is 
formed by condensation of the water vapour in the air, which 
has been derived by evaporation from the surface of the sea, 
lakes, vegetation, soil, etc. Air at a certain temperature 
can only hold a certain amount of water vapour, and this 
amount decreases very rapidly with falling temperature. 
Hence, if saturated air is cooled, some of the water vapour 
which it contains is condensed to form cloud, and, if the 
condensation is continued far enough, rain or snow. Hail is 
formed when a column of air rises very rapidly to great 
heights, as in thunderstorms. The cooling of moist air is the 
only way in which an appreciable amount of precipitation 
can be produced. If the pressure on a mass of air or any 
other gas is lessened, the gas will increase in volume, and in 
doing so will become colder, unless heat is supplied from 
without. At sea-level the air is subjected to the pressure of 
the whole of the atmospheric column above it, equivalent in 
weight to about fifteen pounds per square inch. As we go to 
higher levels and leave part of the atmosphere below us, the 
mass of superincumbent air becomes less, and the pressure falls. 
Hence any sample of air which rises to higher levels in the 
atmosphere will expand and cool, while a sample which 
descends to lower levels will be compressed and warmed. 
That is the reason why, generally speaking, the air is colder the 
higher the level. 

It will be understood from this that very nearly all 



precipitation falls from air which is rising and therefore 
expanding and becoming cooled. From the circumstances 
under which the air rises, precipitation is classified into three 
types Orographic, Cyclonic, and Convectional or Instability 

Orographic precipitation falls where a current of air 
encounters high ground and is forced to rise. The hilly parts 
of Western Britain, standing in the path of the moist south-west 
winds from the Atlantic Ocean, receive a great deal of 
Orographic rain, and are, in fact, the wettest parts of these 
islands. The west coast of Norway is similarly situated ; 
other regions are the eastern end of the Black Sea, the 
mountains of Lebanon, the coast of Honduras in Central 
America, the Western Ghats of India, and especially the 
Khasi Hills of Assam, one of the wettest spots on the globe. 
Orographic rain may be very heavy in warm countries where 
the air contains a great deal of moisture, but in this country 
it is persistent rather than heavy. Much depends on the 
topography ; a long range of hills of uniform height is more 
effective as a rain-maker than an isolated mountain, since, 
unless it is already in an unstable condition, air will always 
go round an object rather than rise above it. Orographic 
precipitation in cold regions frequently falls as snow in winter. 

When we study the distribution of precipitation in a 
mountainous region, such as the Alps, we find that the amount 
is moderate in the lowlands and valleys, and becomes greater 
as we ascend the slopes of the mountains. At a certain height, 
however, termed the " level of maximum precipitation," 
this increase with height ceases, and above this level the 
precipitation becomes less as we ascend. This level depends 
on the temperature and relative humidity of the air over the 
lowlands, and the vertical decrease of temperature ; in the 
Alps it occurs at a height of about 7,000 feet, where the pre- 
cipitation is between two and three times that over the lowlands. 
We can also distinguish the level of greatest rainfall in the 
Alps between 4,000 and 5,500 feet, and the level of greatest 
snowfall in the Alps at about 8,000 feet. The latter must 
not be confused with the snow-line, which is about 2,000 feet 
higher in the Alps ; the level of maximum snowfall depends 
on the winter conditions, while the snow-line is determined 
very largely by the conditions in summer. The relative 



position of the snow-line and the level of maximum snowfall 
are of great importance for the development of glaciers, as we 
shall see in Chapter XVI. 

Cyclonic precipitation falls during the passage of barometric 
depressions, cyclones, and other forms of atmospheric dis- 
turbance which depend on general rather than on local 
conditions. The hurricanes of tropical and sub-tropical 
regions bring torrential rains which often cause widespread 

Fig. 20. Tracks of depressions. 

floods and add to the havoc of the winds, but it is only in 
temperate and sub-polar regions that cyclonic rain forms an 
important part of the total average rainfall. In the eastern 
half of Britain the greater part of the winter rainfall is cyclonic, 
and the same is true of the winter precipitation of all parts 
of the temperate belt except the high ground near the sea. 
The distribution of cyclonic precipitation is largely governed 
by the tracks followed by barometric depressions. Although 
the paths of individual depressions in temperate regions often 
appear to be erratic, it has been found possible to classify 
them into a number of tracks, which are more usually followed 


than the intervening regions (Fig. 20). These tracks have a 
preference for moist areas, especially inland seas such as the 
English Channel, the Baltic, and the Mediterranean, or for 
well-watered plains such as Hungary and Poland. Track I, 
the favourite track at present in all seasons but spring, runs 
from Iceland or the Faroe Islands north-eastward, some 
distance off the coast of Norway, to the Arctic Ocean, or across 
the north of Norway to the White Sea. 

This question of the tracks of depressions is important 
for palaeometeorology, for a considerable degree of permanence 
has been attributed to them. During the Quaternary Ice-Age, 
when the northern tracks were closed by the ice-sheets and the 
glacial anticyclones which occupied them, Track V, which 
runs south of the main glaciated area, was the favourite track. 
Many depressions passed from end to end of the Mediterranean, 
and the rainfall associated with them caused the north of 
Africa to be much moister than at present ; this was a " pluvial 
period " in that region. Track Vb, passing northwards 
across Central Europe, was probably also extensively followed 
during the glacial period, and caused heavy snowfall over the 
south-eastern margin of the ice-sheet. At present it is followed 
chiefly in spring, and is associated with high pressure and low 
temperature to the north-west ; it brings cold spells in Central 

Marsden Manson ( i ) supposes that the cyclone tracks across 
North America have been fixed in their present position 
throughout the whole of geological time, and that the dis- 
tribution of precipitation has always resembled that existing 
at present. He supports this theory by the coincidence that 
the pre-Cambrian glaciations of that continent occurred in 
the present storm belt, but the geological evidence does not 
warrant the generalisation, since North America had an arid 
climate during a large part of geological time. During the 
Eocene period, when the plant-bearing deposits of the Arctic 
Circle were formed, the winds on the west coast of Greenland 
in 70 N. appear to have been mainly south-westerly, indicating 
that the area of lowest pressure and, presumably, of heaviest 
rainfall, lay still farther north, and other evidence will be 
adduced later. It will be an important part of future work to 
lay down the storm tracks of past ages as closely as possible, 
since this will provide a large amount of information as to 



the barometric distribution and the position of the belts of 

The precipitation associated with barometric depressions 
falls as rain or snow according to the temperature at the level 
of condensation. Not infrequently in winter the northern 
half of a depression brings snow, while the southern half brings 
rain. During the Quaternary Ice-Age this distinction between 
the southern and northern halves was probably very pro- 
nounced, and depressions skirting the ice-sheets must have 
caused a large annual snowfall on the ice. As explained in 
" The Evolution of Climate/' this tendency to a maximum 
snowfall near the margin of the ice-sheet appears to have 
played a large part in causing the successive development 
of centres of glaciation more and more to the south-west 
Scandinavia, Scotland, Ireland. 

Convectional or Instability precipitation is typified by the 
hail or heavy rain of thunderstorms. It is due to the warming 
up of the lower air, by contact with ground warmed by the 
sun. When the warming has proceeded far enough, the 
lowest layer of air becomes potentially lighter than the air 
above it (i.e., its temperature is so much higher than that of 
the layer above that even after the expansion and cooling 
consequent on lifting it to the level of the latter it would still 
be warmer). Under these conditions the surface air begins 
to rise through the air above it, at first in thin threads, but 
if the process goes far enough, in thicker columns. The 
first result of this process is the formation of the small cumulus 
clouds so commonly seen in England on a summer afternoon ; 
very often in this country the process goes no farther and the 
clouds die away in the evening without producing rain, 
but under favourable conditions of vertical distribution of 
temperature (and in the presence of sufficient moisture), 
sudden thundery showers or true thunderstorms may result. 
The mechanism of a thunderstorm is complex, and need 
not be discussed ; here it is sufficient to remark that the 
typical thunderstorm is essentially the product of hot, relatively 
calm weather and moist air. The thunderstorms associated 
with " cold fronts " during the passage of depressions have 
a different origin, and the rainfall associated with them comes 
under the heading of cyclonic rain. 

On tropical coasts, the rising of the warmed air over the 


land is facilitated by the onset of the relatively cool sea breeze, 
and the progress inland of the latter is marked by a line of 
thunderstorms. In many parts of the tropics the greater part 
of the rainfall is caused in this way. 

Thunderstorm clouds often extend to a great height, and 
the temperature of the upper part of the cloud may be below 
freezing point. Under these conditions the precipitation 
often takes the form of hail, which is typically associated 
with thunderstorms. It is an important point in the theory 
of the climate of the " warm " periods that the vegetation 
of Europe during the Tertiary period often shows evidence 
of damage by hail, and of torrential rains of the instability 
type, even when the stage reached by the plants shows that 
the deposit was formed quite early in spring. During these 
warm periods the relief was generally low, giving little oro- 
graphical rain, and depressions were probably weak and 
sporadic, but conditions were especially favourable for the 
occurrence of thundery showers, and most of the precipitation 
was of this type. 

The origin of the snow required for the nourishment of 
ice-sheets is a difficult problem. The snowfall of Greenland 
appears to be due almost entirely to the winds from the 
ocean, which blow from the sea on to the ice during the passage 
of intense barometric depressions, but for the far larger ice- 
sheet of Antarctica this explanation is only tenable for the 
margins. W. H. Hobbs (2) considered that the process was 
as follows : the air which flows outward on the surface of the 
ice-sheet must be replaced by inflowing moist air at higher 
levels. These upper air currents are indicated by clouds 
which can be seen passing inland across the coast. In the 
centre of the anticyclone this air descends, and is of course 
warmed by compression, but owing to the intense radiation 
the surface of the ice is intensely cold, colder in fact than the 
air at considerably higher levels (this Antarctic inversion of 
temperature is a well-known phenomenon). The descending 
air is cooled by contact with this intensely cold ice surface, 
and deposits its moisture in the form of small granular masses 
of ice. The formation of an ice-mist of small crystals of ice 
in this way has been observed in Siberia, but apparently the 
precipitation was not sufficient to be termed snow. 

Sir George Simpson (3) showed that this ingenious 


mechanism cannot be the whole story, since the amount of 
snowfall which it would yield would be insignificant. He 
therefore carries the argument a step farther. The moist 
air which flows in at high levels is brought down to the surface 
of the ice by the anticyclonic circulation, and is cooled by 
radiation and by contact with the ice. In this way it becomes 
about as cold as when it crossed the margin of the continent. 
If it becomes colder, some snow will be precipitated in the way 
supposed by Hobbs, but, in any event, it will be almost or 
quite saturated. Owing to local circumstances there are 
irregularities in the temperature distribution ; the coldest 
air tends to spread out on the surface and flow down slopes, 
undercutting and lifting up the air which is less cold. The 
latter is forced to rise again ; it is cooled still further by 
expansion, and consequently snow is formed. This account 
agrees with the fact that by far the larger part of the snowfall 
of the Antarctic continent occurs in blizzards. 

We are now in a position to discuss in general outline the 
distribution of precipitation over the globe. The local 
details are so complex that it is not possible to present the 
distribution adequately on a small-scale map ; this distribution 
is intimately connected with the local geography, and for our 
purpose it is more important to have generalisations applicable 
to any land and sea distribution. In the first place, we may 
distinguish four zones of precipitation in each hemisphere : 

1. The equatorial belt of heavy rainfall. 

2. The sub-tropical dry belt. 

3. The temperate rain belt. 

4. The polar cap of generally light snowfall. 

These four rainfall belts correspond with the four pressure 
belts described in Chapter II., the equatorial belt of low 
pressure, the sub-tropical anticyclonic belt, the temperate 
storm belt, and the polar caps of relatively high pressure. 
The belts of pressure are best developed over the oceans, and 
it is probable that the same is true of the rainfall. The most 
detailed estimate of the zonal distribution of rainfall was made 
by C. E. P. Brooks and T. M. Hunt (4), and the results are 
shown in Fig. 21. 


The rainfall in inches for the different 10 zones of latitude 
are as follows : 

Latitude N. 
Land (in.) . 
Oceans (in.) 

Latitude S. 
Land (in.) . 
Ocean (in.) 

9O-80 80-70 70-60 60-50 50-40 40-30 3O-2O 20- 1 IO-O 

5'8 I2'I 19-3 20-2 23*2 26'6 32'I 56-3 

(5) 8-0 27-2 56-4 51-1 44-2 38-6 48-0 63-3 

0-10 IO-2O 2O-3O 30-40 40-50 50-60 60-70 70-8O 80-90 

60 -I 42-7 26'0 22-2 31-3 38-4 6-9 2'4 (2-0) 

54'3 42'5 36-7 43*4 47'9 37'8 15-0 3*7 

Table 1 1 . Distribution of precipitation according to latitude. 

The rainfall zones are clearly shown in the figures for the 
sea and in the land over the Southern Hemisphere. In the 
Northern Hemisphere the great continental masses of Eurasia 


Fig. 2 1 . Variation of rainfall with latitude. 

and North America are largely beyond the influence of winds 
blowing directly from the oceans, and except near their coast- 
lines they are relatively dry. 

The total amount of precipitation over the whole earth, 
including both land and water, in the course of a year averages 
40-9 inches. This must be equal to the total amount of 
evaporation, since the amount of water vapour and of water 
droplets or ice-crystals in the form of cloud held in the air 
at any time is equivalent to only a few millimetres of rain 
less than one per cent, of the average annual fall. Evaporation 
and precipitation are intimately related, since both depend 
largely on vertical movements in the atmosphere. Rising 
air, as has been pointed out, is responsible for very nearly 
all the condensation of water vapour to form rain, snow, or 


hail, and every ascent of air must be counterbalanced some- 
where by descending air. This is warmed by compression 
as it descends ; its capacity for holding water vapour is 
increased, and it arrives at the surface with a low relative 
humidity. Evaporation takes place chiefly into air which 
has recently descended to the surface from higher levels in 
this way ; it also occurs where air is blowing from a colder 
to a warmer surface and is being warmed by contact with the 
latter. The total amount of precipitation, therefore, is not 
necessarily proportional to the area of the oceans even in low 
latitudes, an assumption which is sometimes made. A 
narrow tropical sea with mountainous shores, across which a 
steady wind is blowing, would be subject to intense evapora- 
tion, but a wind of the same velocity blowing over a stretch 
of ocean a thousand or more miles in length, even if initially 
dry, would at the conclusion of its journey be nearly saturated, 
and would have almost ceased to evaporate. 

Vertical motion of the air, whether caused by winds blowing 
against a range of mountains, by great cyclones or barometric 
depressions, or by local convectional movements, always 
reduces to a question of temperature contrasts between regions 
either remote or near at hand. In particular, the presence of 
ice is a potent factor in causing vertical movement of air. 
Hence a period with warm climates extending over the greater 
part of the world would probably be a period of less evaporation 
and therefore of less total rainfall than a glacial period, in spite 
of the increased amount of water vapour which the warmer 
air can take up ; conversely, during a glacial period, even 
if the temperature is lower, the total rainfall may be increased. 
This was exemplified during the Quaternary Ice-Age, when 
the rainfall over the non-glaciated regions was heavier than 
the present rainfall, and probably much heavier than that of 
any part of the Mesozoic or Tertiary periods. 

We have to take account of the evaporation in another 
way. The biological effectiveness of rainfall depends not 
only on the total fall, but much more on the amount which 
becomes available for plant life. In regions subjected to 
great evaporation the effectiveness of the rainfall is greatly 
diminished, and a hot country with the rainfall of South- 
eastern England would be regarded as dry. The rainfall of 
Jerusalem, for instance, is heavier than that of London, but 


i6 7 

owing to the greater evaporation it is less effective. R. Lang 
(5) expresses the effectiveness of the rainfall as the " Rain- 
factor/ 5 which he obtains by dividing the annual precipitation 
in millimetres by the mean temperature in centigrade degrees, 
and he finds that the character of the soil is closely related 
to this factor in the way shown by the accompanying diagram 
(Fig. 22). According to this diagram, with a mean annual 
temperature of 30 G. (86 F.), desert formations may occur 
with a rainfall as high as 1,200 mm. (48 inches) a year, but if the 
mean temperature is only 10 C. (50 F.), the rainfall cannot 
be more than 400 mm. (16 inches). With a rain-factor 


Salt, Dust 
and Sand 

10 20 

40 60 80 
Annual rainfall in inches. 

100 120 

Fig. 22. Relations of soil to temperature and rainfall. 
After R. Lang. 

between 40 and 60, the deposits are still coloured entirely by 
iron oxides, but the chemical composition and the colour 
vary according to the mean annual air temperature. When 
the latter lies between 32 F. and 54 F., yellow earths are 
formed ; between 54 F. and 68 F., red earths ; and above 
68 F., deep red loams or laterite. The latter, therefore, 
requires a rainfall of at least 800 mm. (32 inches) a year, 
and may be taken as evidence of a fairly moist warm climate. 

When the rain-factor is between 60 and 160, the colour of 
the deposit is due to red or yellow iron oxides and partly to 
black humus, the resulting mixture giving the deposits a brown 
colour (brown earths). With the rain-factor between 100 
and 1 60, the colour is determined entirely by humus, and 


" black earths " result ; above 160 the earth is bleached of all 
colouring matter by the rich vegetation and the heavy rain, 
and the result is a white subsoil surmounted by a deposit of 
pure humus. 

When the mean temperature is below o C. (32 F.), all 
chemical action ceases, and the purely mechanical deposit 
takes the colour of the rock from which it was formed. 
Deposits formed from igneous rocks, or from a mixture of 
rocks of different colours, are generally grey, and the deposits 
formed near the edge of an ice-sheet are often of this colour. 

A more recent table by E. M. Crowther (6) is based on a 
" leaching factor " R 3-3 T, where R is the rainfall in cm. 
and T the mean annual temperature in G. The results, 
based on work in U.S.A., may be summarised as follows : 

Leaching factor Temperature increasing 

above 70 Podsol ; Brown Forest Soils ; Ferruginous 


(Leaching factor 
below 70, rain- 
fall above 70 cm.) Prairie soils. 

Leaching factor Rainfall increasing 

and rainfall both Grey Chestnut Brown Tchernosem 

below 70 desert ; soils semi-desert 

soils soils ; 

Table 12. Relations of soil to temperature and rainfall. 
After E. M. Crowther. 

The biological effectiveness of the precipitation also depends 
on the proportion of it which sinks into the soil. This is 
governed by a number of factors, especially the character of 
the fall and the nature of the soil and vegetation. A per- 
sistent " soaking " rain of moderate intensity is much more 
effective than a torrential downpour which runs quickly off 
the land and floods all the streams, although the actual 
amounts may be the same. A snow-cover which accumulates 
during the winter and melts gradually in spring may be very 
effective. Thick vegetation covering a soft soil checks the 
rate of run-off and allows a larger proportion of the rainfall 
to be absorbed than does hard bare earth. This aspect of 


the rainfall has been brought out by the discussions of the 
desiccation of South Africa (7). In the past fifty years the 
country has been suffering increasingly from drought, but the 
conclusion from expert evidence is that this is not due to an 
actual decrease in the amount of rainfall, but to a change 
in the nature of the soil and vegetation. When South Africa 
was first settled, the country was covered by a rich vegetation, 
the rainfall was steady and persistent, and a large proportion 
of it was absorbed. The effect of over-pasturage has been 
to destroy much of the protective vegetation, and the soil has 
been washed away or trampled hard. The temperature 
contrasts have been increased owing to the heating effect of 
the sun on the patches of bare ground, and the rain now falls 
largely in heavy " instability " showers, including destructive 
thunderstorms. The run-off is proportionally greater, owing 
to the more torrential nature of the fall and the loss of the 
vegetation, so that with nearly the same rainfall the amount 
of water available for use has decreased. The possibility of 
changes of this nature brought about by human activities has 
to be remembered in all discussions of the vexed question of 
" desiccation " in historic times ; in fact a passage in Plato's 
" Critias " suggests that the decadence of Greece may have 
been due to such a change. 

The distribution of the precipitation among the seasons 
is almost as important as the total amount. We have to 
distinguish between regions with their precipitation almost 
equally distributed throughout the year, regions with their 
rainy season in winter and dry summers, and regions with their 
rainy season in summer. The character of the seasonal 
distribution governs the type of vegetation on the one hand, 
while on the other hand it is intimately related to the general 
meteorological regime. This often enables us to derive 
important information as to the general meteorology of a 
period from a study of the plant remains ; for instance, the 
presence of annual growth rings in tree stems may mean a 
seasonal alternation of temperature, but associated with 
evidence of a high degree of warmth, it implies the alternation 
of dry and rainy seasons and a monsoon type of climate. 

The limits of the various rainfall types have been set out 
in great detail by W. Koppen (8). Along the immediate 
neighbourhood of the equator is a belt of heavy rain in all 


seasons. This is the region of the dense forests of the Amazon 
and Congo River basins, and it is known as the " tropical 
rain-forest region." It also includes most of the East Indies 
and the Malay Peninsula, and extends into Cambodia and 
Assam. On low ground the tropical rain-forest belt does 
not usually extend more than 10 on either side of the equator, 
but under favourable conditions it may extend to 20 or even 
to nearly 30 latitude. These outlying areas include the 
eastern coast of Madagascar, the eastern slopes of the Andes, 
and the " everglades " of Southern Florida, another region 
of tropical swamp. 

The soil is usually a rich dark humus. The following 
graphic description is given by E. Warming (9) : " Forest is 
piled upon forest. The trees forming the highest storey have 
tall thick trunks, which are unbranched up to a height of 120 
to 150 feet or more. Beneath them are trees of moderate 
stature with branches not reaching those of the higher tier. 
Beneath these in turn succeed slender thin-stemmed low palms, 
tree-ferns, and shrubs . . . scattered about are huge herbs 
which reach 12 or 15 feet in height. If there still remain 
space available on the ground that is reached by the light, it is 
occupied by dark green ferns, Selaginella, mosses, and similar 
scrophytes. But often the light is too feeble to permit of more 
than a very small number of plants developing on the ground, 
which then may be almost bare of vegetation, with its black 
humus covered only by fallen decaying wet leaves, twigs, and 
remnants of fruits, between which only bizarre saprophytes 
find places . . . but there are hordes of epiphytes clothing 
trunks and branches ... as well as ferns, mosses, and so 
forth. Trees of the forests situate in the cloud-belt of Java 
and the Moluccas are enveloped in a soaking mossy felt, which 
may be thicker than the trunks themselves and imparts to 
them a peculiar dark appearance. . . . Finally, there is 
a wealth of lianas, whose flowers and fruit one can rarely see, 
and whose long, often curiously shaped stems, span the distance 
between soil and tree-crowns, or hang down from the latter 
or partly trail along the ground. . . . The twilight prevailing 
is much less dark than in European beech-forest. All the 
species . . . seem to abhor a vacuum and to combine in 
an endeavour to utilise all the space available." 

Many of the species show protective devices against the 


very heavy showers, especially a smooth cuticle which cannot 
be wetted, drip tips and channelled nerves, but paradoxically 
some plants of the highest storey show xerophytic characters. 
A highly important feature for our purpose is that the trees 
show no annual growth rings. 

Where the contours of the ground are favourable, as in 
Eastern Sumatra, great tropical swamps are formed, which 
appear to reproduce the conditions prevailing during the 
formation of the coal measures, and H. Potoni6 (10) believes 
that the coal measures were in fact formed under similar 
conditions in a very warm rainy climate. The forests which 
formed the coal measures seem to have been similar in many 
respects to the present tropical rain-forests. As described by 
David White (n), the Carboniferous forests showed " rankness 
of terrestrial vegetation ; great size of trees, plants, and leaves 
. . . great size of fronds, and absence of annual rings," while 
" fairly well-developed palisade tissue points towards sunlight." 
The lianas of to-day were represented by " many long slender 
clambering or climbing ferns and fern-like types." 

A. Wegener, in discussing the movements of the poles 
according to his theory of " Continental Drift " (see Chapter 
XIII. ), makes great use of the beds of coal for determining 
the position of the equator in the successive geological periods. 
It should be remarked, however, that although peat is forming 
at present in the tropics in one or two isolated regions, it is rare, 
while in the temperate rain belts peat bogs occur wherever 
there is sufficient moisture. 

Nearer the poles, in temperate latitudes, are other areas 
in which rain or snow falls in sufficient quantities in all seasons. 
These areas include Northern and Eastern North America, 
where they pass into the tropical rain-forest area, all Northern 
and Central Europe, and a large part of Asia ; in the Southern 
Hemisphere they are limited to relatively small areas in Chile, 
South-eastern Australia, and New Zealand. In Europe the 
area occupied by these temperate rains was the site of great 
peat formation during the post-glacial period, and certain 
coal beds of earlier geological periods are attributed by 
Wegener to the temperate rain belt. 

On either side of the equatorial " rain-forest " belt, and 
extensively developed on the eastern sides of the continents, is 
the monsoon or summer rain region, best known from its 


occurrence in India and China. This type of distribution 
is due to the extensive alternate heating and cooling of the 
interiors of large continents and the consequent alternation 
of monsoon winds ; it occurs on either side of the tropical 
rain-forest belt in South America and West Africa, and is very 
extensively developed in Eastern Africa and in Southern and 
Eastern Asia ; it is also found in the north and east of 
Australia. The type of vegetation associated with it is the 
savannah or meadow land, passing into open forest with 
increasing rainfall. Where the rainfall is especially heavy and 
the temperature steadily high, it may give dense tropical 

On their poleward sides, the summer rain regions in the 
western and central parts of the continents usually pass into 
deserts, which are characterised by a slight and irregular 
rainfall, many months sometimes passing without even a 
shower. In parts of the South American desert it has probably 
not rained for centuries. Such plants as there are show 
special devices to prevent the loss of water, but in many 
deserts the ground is entirely bare of plants. Among animals, 
one of the most characteristic is the lung-fish (Ceratodus] , which 
is adapted to breathe either air or water, and can lemain 
dried up for long periods. This form, which is still living in 
Australia, has persisted since the Permian (allied forms are 
known since the Devonian), and affords some evidence of 
the continuity of desert conditions throughout a large part 
of geological time. The best geological evidence of arid 
climates is lithological desert sandstones of rounded and 
polished grains usually red in colour, " dreikanter," and other 
wind-eroded rocks, deposits of gypsum and of salt ; these 
abound in many horizons and seem to suggest that the desert 
belts were greatly expanded in the past. This was probably 
true, but we have to remember that before land plants reached 
their present degree of specialisation, large areas which would 
now be habitable by plants must have been bare rock, subject 
to aerial denudation, so that " desert " sandstones do not 
necessarily imply a rainfall as small as that of modern deserts. 

On the poleward sides of the deserts the rainfall increases 
again, but falls mainly in winter, while the summers are hot 
and dry. The best known example of this climate is the 
Mediterranean region, whence the type is often known as the 


Mediterranean. It is found also in California and in limited 
regions in Chile, near Cape Town, and in Southern Australia ; 
it is thus practically limited to the belts between latitudes 30 
and 45 and, except in Mesopotamia, it never extends far 
from the sea. Eastward it usually passes into a semi-arid 
or arid climate. It gives a peculiar type of vegetation adapted 
to resist the drought of summer and to maintain its moisture 
from considerable depths, for example, the vine. 

The Mediterranean type of climate appears to have per- 
sisted in South-east Europe during a large part of the Tertiary 
period (12), though with variations in the total amount of 
rainfall. The region was an archipelago of small islands ; 
in the Early Eocene it had a rather moist climate with some 
rain in all seasons, but in the Middle Eocene the entire absence 
of a land flora probably indicates a semi-arid climate. In 
the Oligocene there were well-marked dry and rainy seasons, 
and in the Early Pliocene the climate was again warm and 
rather moist. 

On its polar side, the Mediterranean climate passes into 
the temperate rain belts, which, except in the interior of the 
great continents of Eurasia and North America, have a 
sufficiency of rain at all seasons, and are occupied mainly by 
forests of conifers or deciduous trees, which in the more 
equatorial parts of the belts include some sub-tropical species. 
Nearer the poles the winter is severe, with a persistent snow- 
cover, and the summer is short ; this is the " boreal " climate. 
Near the coast there is abundant rain at all seasons ; in the 
interior the winter is usually dry, but the annual variation 
of temperature in all cases enforces a period of inactivity in 
plant growth. This zone is the peculiar home of peat bogs, 
which require a rainfall of at least 40 inches a year and a 
mean temperature above 32 F. The great outbursts of peat 
formation in the " Atlantic " and " sub- Atlantic " periods 
of post-glacial time gives us a measure of the raininess of these 

The distribution of the belts of rainfall is closely related 
to the distribution of the belts of pressure and wind described 
in Chapter II., and the temperature belts described in Chapter 
VIII., and we may set out the general succession of climatic 
belts in the way shown in Table 13. The first column gives 
the average latitude in which the different belts are found 




: : ' ': <U : 

cS - !.S - i "8 ': 

" ^ . 

fe S : rt S fc ' 



& u iS '& & U 

g jJ O j^rf O 4 ^* : ^ 


. *4J 

_fi m : *j _fiCQ ...n : bT 
4J C : ^ . +3 4J jij : >^ 

d jg 

O *c3 o ^ rr^* : ^ u^ ^ "3 


^Pi ^;Q ;WM I^Q OTrt :w 

: </a ' : : : c/5 : 

<U ' C^ . 

: c/5 

j "g ; J4J j "2 :^ 

! i 

: *^- 

i 8 : i iwi g j:| 



Polar East V 

\ > \ S S3 i'% SS IS; ^ i^ 

: ^ j^^Ojc-o ;^3; ^g jjg 

: rn : ^3 "Ti : 0tf : ^ !H 
W ^O -M O -*- :cj: PH ;<^3 

S : 'fi 2 : r^ 52 : *S : r^ ! "S 
p C3 i rH C^j Q.,' -t i w 

. ^ ; 0^ V : ^^ V : : S3 j ^ 





i i 

; <"} 

C^ ; ^3 f~] ! EH J f"j ; ^j ; f| | C] 


: -<-> 
: ^ 

S _o "G : ^ S 3 : JD ' 3 J 3 


: O 

^2 ^^KSi^l ;;3; * 

: (T, i : C/D ^-4 W J C/} : C/D : C/D J C/3 


: : : S : : 1 





'"*"* '' ; 4-> ^ f 








' p | ' ' p ^^ 

C2 ', ' Cj ' T3 






i s i ? i^ i ^ 



JH 4-J *-> : *S 4-* : ^ 


: O ^ : O ; o 'S : O 
: Z M : W : 0) PQ i & 


+"* j-J O 




1 i 1 3 



S N S ^ 



^ 1 ^ ^ 




^f* "" r{ r{ 

fe CJ O 3 C3 S 

!SN H c^N co 








in the two hemispheres, while the second column gives the 
" astronomical " zones of the old school text-books, which are 
limited by the tropics and the polar circles. The third 
column shows the belts of temperature according to Supan, 
who limited his " hot belt " by the mean annual isotherm 
of 20 C. (68 F.), which approximately fixes the polar limit 
of palms, and his cold caps by the isotherms of 10 G. (50 F.) 
for the warmest month, which forms the limit of growth of 
cereals and forest trees. It will be noticed that the cold cap 
extends into much lower latitudes in the Southern than in 
the Northern Hemisphere ; this is due partly to the influence 
of the great Antarctic ice-cap and the ice-laden Southern 
Ocean, and partly to the smaller annual range of temperature 
over the oceanic regions of the Southern Hemisphere than over 
the continental regions of the Northern Hemisphere. 

The rainfall belts have a close relation to the wind systems. 
The equatorial rain belt is limited to the Doldrums and part 
of the region of the South-east Trades. The dry belts include 
most of the trade wind regions, the whole of the sub-tropical 
calm belts, and extend a short way into the domain of the 
westerly winds, where they give place to the temperate rain 
belts, which in turn are limited on their poleward sides by the 
polar east winds. We have seen reason to believe that when 
the temperature-difference between equatorial and polar 
regions was below a certain critical value, the polar caps of 
east winds would be suppressed and the westerly winds would 
extend almost to the poles. This would result in a spreading 
out of all the other zones into higher latitudes, and would carry 
the poleward margins of the sub-tropical dry belts into the 
regions at present occupied by the temperate rain belts, while 
the latter moved northward to occupy the polar " cold caps," 
now characterised by tundras, glaciers, and in Greenland and 
the Antarctic by " the climate of eternal frost." 


(1) MANSON, MARSDEN. " The physical and geological traces of the cyclone 

belt across North America." Washington, Monthly Weather Review, 52, 
19124, p. 102. 

(2) HOBBS, W. H. "Characteristics of existing glaciers." New York, 1911. 

(3) BRITISH ANTARCTIC EXPEDITION, 1910-1913. Meteorology, vol. i. Discussion, 

by G. C. SIMPSON. Calcutta, 1919. 

(4) BROOKS, C. E. P., and T. M. HUNT. " The zonal distribution of rainfall 

over the earth." London, Mem. R. meteor. Soc. y 3, no. 28, 1930. 


(5) LANG, R. " Verwitterung und Bodenbildung als Einfuhrung in die 

Bodenkunde." Stuttgart, 1920. 

(6) CROWTHER, E. M. " The relationship of climatic and geological factors 

to the composition of soil clay and the distribution of soil types." London, 
Proc. R. Soc., B, 107, 1930, p. i. 

(7) UNION OF SOUTH AFRICA. " Report from the Select Committee on 

Droughts, Rainfall, and Soil Erosion." Cape Town, 1914. " Final 
report of the Drought Investigation Commission, October 1923." Cape 
Town, 1923. 

(8) KOPPEN, W. " Die Klimate der Erde." Berlin und Leipzig, 1923. 

(9) WARMING, E. " (Ecology of plants." Oxford, 1909. 

(10) POTONIE, H. " Die Entstehung der Steinkohle und der Kaustobiolithe." 

5 Auf. Berlin, 1910. 
(n) WHITE, D. "Upper Palaeozoic climate as indicated by fossil plants." 

Sci. Mori., New York, 20, 1925, p. 465. 
(12) KERNER, F. " Bauxite und Braunkohlen als Wertmesser der Tertiar- 

klimate in Dalmatien." Wien, 1921. 


IT has long been evident that the succession of climates 
in the various geological periods has not been haphazard, 
but has followed a certain ordered sequence. The last 
climatic episode on the grand scale has been Quaternary 
glaciation. Looking back beyond that, we see a long succession 
of genial climates in the Tertiary and Mesozoic, until we come 
to the great Upper Carboniferous glaciation in the late 
Palaeozoic. Beyond that again is another long period of 
mainly warm climates Lower Carboniferous, Devonian, 
Silurian, Ordovician, and Upper Cambrian, bringing us to 
another great glaciation at the close of the Proterozoic, 
immediately preceding and perhaps extending into the 
Cambrian period. Beyond that again, and almost lost in 
the mists of antiquity, deposits of a still earlier glaciation have 
been recognised in South Africa, Australia, North America, 
and perhaps in Scotland. 

The systematic nature of these occurrences is made more 
obvious when we consider their probable absolute ages in 
years. The Quaternary glaciation is recent, it began perhaps 
one million years ago. From the evidence of the radio-active 
rocks (see Appendix I.), it is calculated that the upper part of 
the Carboniferous system is about 260 million years old. 
The base of the Cambrian is placed at 500 million years. 
The exact age of the first of the four great glaciations is not 
known, but a fair estimate would be 700 to 800 million years. 
Thus the great glaciations seem to have occurred at almost 
regular intervals of a quarter of a thousand million years. 
The last two of the long genial intervals, and for all we know, 
the first also, have been interrupted by minor deteriorations 
of climate, as in the Silurian and the Cretaceous to Lower 
Eocene, which produced local valley and sometimes piedmont 
glaciers but not regional ice-sheets, suggesting a shorter cycle 
superposed on the longer one. 
The same ordered sequence has been observed in the 


evolution of the earth's surface features, where it has been 
termed " the rhythm of geological time. 3 ' As was remarked 
in the Introduction, there have been long periods in which 
the earth's crust was at rest, while the denuding agencies 
gradually lowered the surface almost to a uniform plain and 
the waves of the sea bit deeply into the continents. Alternating 
with these have been relatively short periods of intense dis- 
turbance during which the earth's surface was thrown into 
great folds and ridges, when the mountain ranges which form 
the articulated skeletons of the continents were brought into 
being. The greatest periods of mountain-formation occur 
in close relation with the greatest periods of glaciation ; thus 
the Alpine period of folding in the Tertiary preceded the 
Quaternary Ice-Age ; the Hercynian folding in the Upper 

Lower Upper Ccstedoruaft HcrcyrAiar\ Cretaceous Alpirxe 

Prolcro}0ic u) Prolcro*oic (Silurian) (Upper -Eocene (Quaternary) 

' Carboniferous) ' 

75 O 5OO E5O 

Millions of years 
airx bu'ildirxq (full line) ar\d cjlaclaJlorv (brokers I'mc) Schematic 

Fig. 23. Mountain-building and glaciation (schematic). 

Carboniferous preceded the Upper Carboniferous glaciation. 
It is known that there was a period of great disturbance and 
mountain-building preceding the Cambrian period, and 
another, lower in the Proterozoic, probably preceded the 
first of the four great glaciations, the deposits of which, 
in Australia at least, rest on great outflows of lava. The 
minor cold period of the Silurian was also associated with 
a period of folding and mountain-formation, the " Caledonian," 
which, however, did not reach the intensity of the Hercynian 
and Alpine foldings. Thus we can represent the variations 
of mountain-building activity and of climate during geological 
time as a series of waves (Fig. 23) in which the long troughs 
represent the periods of stability and genial climate, the sharp 
crests the periods of mountain-building and climatic stress. 

The last two ice-ages at least were not synchronous with 
the maximum of mountain-formation, but followed them after 
some millions of years, as has been indicated in the diagram. 


This lag has been attributed to two causes. After a long 
quiescent warm period the whole mass of the oceans is warm, 
and has to be cooled down before general glaciation can begin. 
This process may occupy thousands of years, and smooth out 
climatic fluctuations within an ice age, but could not cause a 
lag of millions of years. The second cause is that mountain 
ranges are first elevated as smooth domes, which are worn into 
irregular contours by the ordinary processes of erosion. The 
lightening of load caused by the removal of this eroded material 
causes further isostatic elevation and a greater effective height. 
A. Wagner (i) has put forward a third explanation. The 
steady warming of the earth's crust by radio-activity is much 
greater than the normal escape of earth-heat at the surface 
so that the crust becomes continually hotter and more plastic. 
This allows folding, mountain-building and volcanic out- 
breaks, in which the accumulated earth-heat is liberated. 
At present earth-heat raises the mean temperature of the 
earth by only 0-3 F., but Wagner thinks that during the 
mountain-building of the first half of the Tertiary this figure 
may have been nearer 10 F., more in high than in low 
latitudes. This radio-active heat warms the earth's crust, 
and melts the base of any nascent glaciers. The latter flow 
rapidly to lower levels where they are dissipated. 

After the end of the main epoch of mountain building the 
crust becomes solid and quiescent and cools again. This allows 
ice to freeze to the ground and so pile up great ice- 
sheets. When the ice reaches a certain thickness the vertical 
temperature gradient in the upper part of the crust increases 
(at cost of lower layers) and melts the bottom of the ice-sheet, 
causing it to flow out to the warmer periphery where it melts. 
This is the phase of maximum extension of the ice-sheets. 
After the ice-sheets have melted back the earth-flow decreases 
again and the cycle recommences. In this way Wagner 
accounts not only for the lag of glaciation behind mountain- 
building but also for the succession of glacial and inter-glacial 
periods. The theory is mentioned here for the sake of com- 
pleteness, but it seems improbable that the surface temperatures 
can have fluctuated so greatly solely because of earth-heat, 
especially during the course of the Quaternary Ice-Age. 
H, Jeffreys (2) states definitely that the surface temperature 
of the earth must have been almost wholly maintained 


by solar radiation practically ever since it became solid at 
the surface, and certainly throughout geological time. 
Conduction from the interior is by comparison quite un- 

The dates of the pre-Cambrian glaciations are still uncertain, 
but have been provisionally put at 500, 1,000 and probably 
1,500 million years ago, with more doubtful occurrences 
round 600 or 800 million years ago. The duration of a period 
of folding and mountain-building is about 50 to 80 million 
years, while that of an ice-age is much shorter, hence the 
peaks representing mountain building have been shown 
broader than those representing glaciation. 

As G. Manley (3) pointed out, the problems of why the 
Quaternary glaciation began just when it did, and of why 
it was interrupted by interglacial periods, is the same, on a 
larger scale, as the problem of the fluctuations of the glaciers 
during the historical period. So far as is known, these have 
been broadly similar over the whole world. Especially 
during the past 100 years glaciers have been in rapid retreat 
in both hemispheres alike. The causes of these minor 
advances and retreats of the glaciers are still unknown. 

The close parallelism between mountain-building and 
climate has naturally attracted the attention of geologists. 
For example, in 1899 J- Le Conte (4) pointed out that the 
Quaternary glaciation was preceded by a period of " almost 
universal continental elevation and enlargement," and that 
the Quaternary ice-sheets developed on the surfaces of the 
plateau so formed. The greatest exponent of the importance 
of elevation, however, has been W. Ramsay (5), who, under the 
title " Orogenesis and Climate," attempted a complete 
solution of the climatic problem on these lines. 

In the second paper (5), Ramsay summarises his theory of 
the climatic importance of elevation : " It is well known that 
a climate is modified by the relief and the elevation of a region, 
so that with increasing altitude it gradually becomes cooler 
and even glacial. Further, mountains and highlands exercise 
a great effect as condensers of precipitation, as boundaries of 
climate, etc. But meteorologists and geologists reflect less 
often that the relief of the continents influences, not only the 
local or regional climate, but the whole economy of the 
calories which the sun supplies." 


We have seen in Chapter VI. that the atmosphere, owing 
to the selective absorption which it exercises on radiation of 
different wave-lengths, acts like the glass of a greenhouse in 
raising the temperature of the earth's surface. Ramsay 
points out that this action takes place under most favourable 
conditions over extensive low plains where the air mantle is 
thickest and most dense. Over the mountains and highlands, 
where the air is thinner and less dense, the loss of heat by 
terrestrial radiation escaping to space is greater, " the lofty 
parts of the continents can be regarded as holes in the glass. 
They not only chill the place just beneath them, but more or 
less the whole hotbed." 

Loss of heat also takes place owing to the vertical movements 
which high ground introduces into the atmosphere. " The 
air currents, passing over mountains, high coasts, and other 
elevations in their way, are forced to rise, and at the higher 
position their loss of heat is greater than if they had flowed 
at a greater level. 55 It may also be remarked, though this 
consideration is not mentioned by Ramsay, that high ground 
favours the formation of cloud which, as shown in Chapter 
VII., lowers the temperature owing to its reflection of solar 
radiation. But the most important effect of high ground, 
according to Ramsay, is that it serves as a gathering ground 
for glaciers. High land is necessary for glaciation, and high 
land in rather high latitudes is especially favourable for the 
occurrence of a general cold period. " If there only exist 
high enough islands and continents, ice-caps will appear, 
extend their glaciers down to the sea, and send out their 
armadas of icebergs. To melt them, enormous quantities 
of the heat reserve of the sea will be consumed. Cold water 
forms extensive superficial layers, and gradually fills the depth 
of the ocean right to the equator. 55 As the ice-caps grow, 
more and more water is bound up in them ; this water is 
taken from the sea and gradually lowers the sea-level, thus 
accentuating the elevation of the land. Antevs (6) calculates 
that during the Quaternary glaciation the average elevation 
of the land above the sea was increased by more than 300 
feet owing to the locking up of water in the form of land ice. 

Thus Ramsay has built up a very complete qualitative 
theory of geological climates on the basis of changes of elevation 
alone. His theory contains a very considerable amount of 


truth ; he has realised the importance of vertical motion in 
the atmosphere for the loss of heat by radiation, and he has 
also introduced the effect of changes in the general temperature 
of the ocean. On the other hand, I think he has overestimated 
the importance of the " holes in the glass," since at the present 
day only relatively insignificant areas of land are high enough 
to make an appeciable difference to the proportion of their 
radiation which is absorbed by the atmosphere. The loss of 
heat by radiation from air-masses forced to rise in order to 
pass over high ground is no doubt appreciable, but the amount 
of air raised in this way must be much less than that raised 
in cyclones, thunderstorms, convection currents, and other 
phenomena of atmospheric instability. In fact, as pointed 
out in Chapter IX., air, unless it is already unstable, much 
prefers going round a mountain to going over it. One would 
be inclined to say that the " polar front " in the North Atlantic 
is responsible for more vertical air movement than the whole 
of the mountain regions of the globe put together, but there 
are no figures available for a quantitative estimate. On the 
other hand, we have seen in Chapter VI. that the loss of 
radiation by reflection from ice-sheets, at the maximum of the 
Quaternary Ice-Age, would suffice to cool the whole earth 
by at least 4 F. 

The weakness of Ramsay's theory, as of so many other 
theories of climatic change, is the lack of a quantitative basis. 
Let us examine the various effects of elevation which he 
postulates and attempt to evaluate them numerically. The 
climatic effects of high ground are, as we have seen, complex. 
The most obvious effect is the decrease of mean temperature 
with height, at the rate of about 0*3 F. per 100 feet. This 
is due to the fact that air which rises from a low level to a 
higher level expands and cools by expansion. But when the 
air descends again it is warmed by compression, so that the 
direct effect of high ground in causing the air to cool by 
expansion is purely local, and cannot influence the temperature 
of the sea surface or of the lowland plains. The effects which 
we have to consider are those which cause a net loss of heat 
from the earth's surface as a whole. These include : 

i. Radiation to space from the high ground which is not 
absorbed by the air. 


2. Reflection of solar radiation from the surface of clouds 

(not postulated by Ramsay). 

3. Cooling power of the surfaces of ice and snow. 

4. Loss of heat owing to increased evaporation. 

Reasons have already been given why the loss of heat under 
the heading i is probably negligible, and this factor will not 
be considered further. 

2. Except in winter in high latitudes, clouds lower the mean 
temperature by reflecting the rays of the sun back to space. 
Where high ground forces the air to rise above the level of 
condensation, clouds will be formed, and the local temperature 
will be decreased. The level of condensation varies according 
to the humidity of the air, but 5,000 feet would be a fair 
estimate of the average level over the land. The air above 
the sea and the lowlands is not by any means free from cloud, 
but it is noticeable that there is a marked increase in the 
cloudiness in the neighbourhood of mountain ranges, and 
especially on the windward sides, which is not entirely com- 
pensated by the somewhat clearer skies on the leeward sides. 
Let us say that the net effect of land above 5,000 feet is to 
increase the local cloudiness by three-tenths of the sky. Now 
we find (7) that about 12 J per cent, of the land, or 3-6 per 
cent, of the whole surface of the earth, is at present above 
5,000 feet in height. The effect of this high ground is therefore 
to increase the cloudiness, expressed as an average over the 
whole earth, by one per cent, or o- 1 tenth of the sky. We saw 
in Chapter VII. that an increase of one-tenth in the mean 
cloudiness results in a decrease of the mean temperature by 
6 F. The effect of the introduction of the present topo- 
graphy into a nearly base-levelled world would therefore be 
to decrease the mean temperature by about 0*6 F. owing to 
the increase of cloudiness over the mountain ranges. 

During the Quaternary, the average elevation of the land 
was considerably greater than it is now ; we do not know 
exactly how much, so we must guess. Let us put the elevation 
at 3,500 feet instead of the present average of 2,500 feet. 
Further, let us suppose that the present elevations were all 
increased in the same proportion, so that instead of there being 
12^ per cent, of the land above 5,000 feet there was the same 
amount above 7,000 feet. From the data given by de Martonne 


(7) we then find that the area above 5,000 feet would have been 
21 per cent, instead of 12 J per cent. The resulting increase 
of cloudiness would have averaged i 8 per cent, instead of 
i per cent, over the whole earth, giving an average cooling 
of i o F. compared with a warm period or o 4 F. compared 
with the present. 

3. The cooling power of the surface of ice and snow is difficult 
to evaluate. The mean temperature of the earth's surface 
reduced to sea-level is at present 59 F. 5 varying from below 
freezing point in high latitudes to above 80 F. in parts of the 
tropics. The " snow-line " accordingly varies from sea-level 
to above 16,000 feet. But this low snow-line in high latitudes 
is really due to the presence of the ice-sheets and floating 
ice ; the glaciers creeping down the sides of the hills have 
brought the snow-line down with them. The cooling power 
of these ice-surfaces belongs partly to the account of elevation, 
but partly to other causes, such as increased continentality, 
shutting out of ocean currents, volcanic action. What we 
are seeking for here is the direct effect of elevation alone, apart 
from the other geographical factors which are connected with 
elevation. Let us put it that, starting with a warm period 
in which there was no ice, a burst of mountain-building 
rapidly raises the average level of the land to 2,500 feet ; 
we require to find the area brought above the snow-line. In 
order to do this it is necessary to make some assumptions as 
to the distribution of temperature and land. We will suppose 
that during the warm period preceding the elevation the mean 
temperature over the land areas is 32 F. at the poles and 
83 F. at the equator ; further, that from the poles to latitude 
45 N. and S. it rises at a uniform rate of 8 F. in each 10 of 
latitude, while from 45 to the equator the mean temperature is 

33 F.+5O cos <, where </> is the latitude. 

Now, suppose that the whole area is elevated in such a way 
that the ratio of land to sea and the percentage areas of land 
above different heights are the same in all latitudes, having 
everywhere the present average value for the whole world, 
the average elevation being 2,500 feet. Then a rough cal- 
culation shows that rather more than three per cent, of the 
land surface, or about one per cent, of the total surface of the 
earth, would be brought above the snow-line, more than half 


this area being between latitude 70 and the poles. The 
cooling power of ice compared with that of unglaciated land 
varies according to the latitude ; from the data given in 
Chapter VIII. it appears that if a land surface in high latitudes 
which was formerly bare of ice have one per cent, of its area 
ice-covered, the mean annual temperature will be lowered by 
about 0-3 F. The ice-covered area is not limited to the area 
initially raised above the snow-line, since valley and piedmont 
glaciers can descend to low levels. The development of 
great ice-sheets probably requires a general cooling of the seas 
in addition to simple elevation, but we may take the areas of 
mountain glaciers below the snow-line as equal to the areas 
above the snow-line. This gives us for the total cooling, 
averaged over the whole earth, due to the development of 
snowfields and glaciers as a result of elevation, an amount of 
0-6 F. 

When the calculation is repeated, supposing the average 
elevation of the continents to be raised to 3, 500 feet, the heights 
of all parts being increased proportionally, the cooling due to 
the snow and ice surfaces, when averaged over the whole 
earth, is rather more than doubled, becoming i'3F. If, 
further, we suppose that the land is concentrated in high 
latitudes, forming two continents extending from the poles 
to about latitude 50, the cooling becomes still greater, ex- 
ceeding 6 F. when averaged over the whole earth. Since 
during the Quaternary Ice-Age the elevation was greatest in 
high latitudes we may assume an intermediate figure, say 

4. The last term to be discussed is the loss of heat due 
to increased evaporation. During the great ice-ages the 
total precipitation was probably more than at present ; 
during the warm periods it was very much less. The 
actual figures can only be guessed at, but judging from 
the widespread aridity of periods like the Triassic, we may 
not be far out if we suppose that the average precipitation 
during the warm periods was about half that during the ice- 
ages. Since the amount of water held in the air as vapour 
or clouds at any time is equivalent to only a fraction of an 
inch of rain, the precipitation gives us a measure of the 
evaporation, so that the evaporation also during the ice-ages 
may have been twice that during the warm periods. Most 


of the evaporation takes place from the surface of the sea. 
Evaporation cools the evaporating surface, and an increase 
in the elevation of the land, by increasing the evaporation, 
lowers the temperature of the sea. 

The loss of heat due to evaporation at present averages 
about 100 gram calories per square centimetre per day over 
the oceans, or perhaps 90 gram calories over the land and sea 
together. The loss of this amount of heat would lower the 
mean temperature by about 15 F., so that the mean tempera- 
ture during the ice-ages may have been lowered by as much 
as 8 F. compared with that of the warm periods from this 
cause alone. But we cannot attribute all this amount directly 
to elevation. As explained in the preceding chapter, the 
greater part of the world's precipitation is probably due to the 
" polar fronts." During the warm periods the polar fronts 
were not developed, and the greater part of the rain fell in 
instability showers of a thundery nature. Such showers, 
though they are sometimes very heavy, are extremely local, 
and do not deliver such a large total of rain as the extensive 
rain areas associated with cyclonic depressions. Much of 
the difference between the total precipitation of the warm 
periods and that of the ice-ages was probably due to this 
cause, and not directly to the differences of elevation. Let 
us put down half the difference to the account of the polar 
fronts and half to the elevation. Then we have an elevation 
of 3,500 feet, resulting in a cooling of 4 F. due to increased 

We can now sum up our three guesses. Comparing an 
ice-age with a warm period, we have : 

Cooling due to increased cloudiness . . . . i F. 
Cooling due to area above the snow-line . . 3 F. 
Cooling due to increased evaporation ... 4 F. 

The total cooling due to the elevation alone is about 8 F., 
and from the difficulties of the calculation may be anything 
from 5 F. to 10 F., but is not likely to be outside these limits. 
Evidently, elevation is by itself and directly an important 
factor in climatic changes, but it cannot account entirely for 
the changes of some 30 F. in the mean temperature of the 
Arctic lands at sea-level, or of the surface of the Arctic Ocean. 
The truth is that mountain-building is so intimately 


associated with changes in the land and sea distribution 
that it cannot really be considered alone. When the con- 
tinents are thrown into folds and ridges, so too are the ocean 
floors, at least near the continents. If the continents are 
more elevated, the oceans are deeper. The waters retire 
into these submarine deeps and expose large areas of the 
continental shelves as new land ; this at once increases the 
continentality and limits the ocean currents. At the same 
time volcanoes break out in the mountain ranges, and by 
ejecting dust into the atmosphere add to the general cooling. 
The polar oceans, cooled below their freezing point, develop 
floating ice-caps, while ice-sheets spread from the mountains 
over the surfaces of the continents. The temperature difference 
between low and high latitudes passes the critical point for 
the atmospheric circulation, and the polar front appears. 
The oceanic circulation is radically changed. The whole 
process is in fact essentially cumulative, and the final lowering 
of temperature is out of all proportion to the small initial 
causes. That is why the whole problem of climatic changes 
is so baffling. 

The time has therefore come to gather up the scattered 
threads which we have been patiently disentangling in the 
preceding chapters, and to attempt to weave them into a 
tapestry which will give us a connected picture of the 
mechanism of climatic changes. As ultimate causes, we 
have a choice between variations of solar radiation, sunspots, 
astronomical changes, ocean currents, continentality, 
mountain-building, volcanic dust, and carbon dioxide, though 
the possibilities of the last-named are limited. As auxiliary 
causes, we have variations of water vapour, of cloudiness, of 
the wind circulation, and, most important of all, of floating 
ice surely enough, between them, to give us sufficient material 
for our purpose. 

Let us start with the conditions of the present day and 
try to forecast the meteorological changes during the next 
quarter of a thousand million years, which, I believe, is the 
longest-range forecast ever attempted. We are still living 
in an ice-age, though not at its maximum ; whether, before 
the forces of mountain-building which began their activity 
in the late Cretaceous have worn themselves out, there will 
be a return of the ice-sheets in Europe and North America, 


we cannot say, so we will skip a million years or so and begin 
our forecast with the period when the subterranean fires are 
banked and the earth begins again to settle down. The 
Arctic Ocean is still filled with ice, and the summits of the 
high mountains are above the snow-line. We have our 
regular succession of storms born on the edge of the polar 
front and travelling eastward into Europe altogether, the 
weather is pretty much as we know it to-day (unless by that 
time the problem of the artificial control of weather has been 
solved who can foretell the triumphs of science?). 

When the mountain-building forces cease to repair the 
ravages of denudation, the average level of the land begins 
to fall steadily. The process is most rapid in the high 
mountains, where frost is still active, but everywhere the 
rivers are carrying sediment into the oceans and the waves 
are grinding into the land. The first noticeable change is the 
gradual disappearance of the mountain glaciers as their 
gathering grounds are worn down beneath the snow-line ; 
the ice-sheets of Greenland and Antarctica still persist, but 
are beginning to wane. Bering Strait has been widened and 
deepened a little ; a small warm current penetrates through 
it into the Arctic Ocean for longer and longer periods in each 
summer, and finally succeeds in keeping clear of ice a " bridge- 
head " on the northern side, and in maintaining its flow 
throughout the year. The Atlantic gap is a little wider-- 
Iceland is smaller and the southern end of Greenland has 
retreated a short distance northward, while at the same time 
the gap between Newfoundland and Labrador has widened, 
allowing much of the ice and cold water of the Labrador 
Current to flow directly down the coast of America, instead 
of mingling with the waters of the Gulf Stream off the Grand 
Banks. Thus the water of the Gulf Stream Drift is a little 
warmer when it reaches the Arctic Ocean, and Spitsbergen 
is now ice-free throughout the year. Attacked on two sides, 
the Arctic ice-cap melts farther and farther back each summer, 
and the Palaeocrystic ice begins to disintegrate. Finally, 
there comes a succession of years in which the sun's radiation 
is unusually powerful the solar constant remains steadily at 
2 o calories per square centimetre per minute and one 
summer the Arctic ice-cap breaks up completely and disappears. 
That summer the polar east winds are greatly weakened and 


the " polar front " hardly develops ; the Azores anticyclone 
extends persistently across the British Isles and Western 
Europe almost to the Urals ; hardly any cyclonic depressions 
occur south of the Arctic Circle, and the summer is almost 
rainless. The cold East Greenland Current carries a little 
ice in spring, but this fails to reach Cape Farewell, and the 
west coast of Greenland is bathed by a warm branch of the 
Gulf Stream instead of by cold Arctic water. The Greenland 
ice-sheet makes a record retreat, but the level of the sea is 
rising still more rapidly, and the ice-edge still reaches the sea 
in many places. 

The Arctic Ocean remains open far into the following 
winter, and when it finally freezes, the ice is thin and easily 
scattered ; it begins to break up quite early in spring, so 
that by the middle of summer it has completely disappeared. 
The " semi-glacial " condition has been reached in the 
Northern Hemisphere. In the following years the ocean 
becomes steadily warmer ; the Arctic ice-cap, and with it 
the polar east winds, the " polar front," and the Atlantic 
cyclones develop later and later each winter and break up and 
disappear earlier and earlier each spring, until finally there 
comes a winter in which the ice-cap does not form at all, and 
even in the middle of winter the Atlantic Ocean is almost 
free of storms. The air in high latitudes is not cold enough 
for its greater density to counterbalance the relatively warm 
and light polar stratosphere, and west winds prevail everywhere 
outside the tropics, with, at the surface, a component towards 
the poles. The late summers in the British Isles and Western 
Europe are now intensely hot and dry ; day after day the 
skies are almost clear of cloud, while even the slight hindrance 
to the sun's rays offered by volcanic dust is absent, and a 
powerful sun warms the surface of the land and of the seas. 
The temperature becomes very high, exceeding 100 F. on 
several days at midsummer ; owing to the high temperature 
the air contains a large amount of water vapour which 
absorbs the terrestrial radiation. Part of this radiation is 
given back to the earth at night, and in spite of the clear skies 
the nights are not cool. (In the dry summer of 1921, which 
made some approach to these conditions, the mean daily 
minimum temperature at Kew Observatory in July was 
nearly 5 F. higher than the mean daily minimum during the 


much cloudier month of July 1924.) There are a number 
of thunderstorms in spring, but by the end of June the higher 
layers of the troposphere have become warmed up to such an 
extent by this absorption of terrestrial radiation that the air is 
stable, and in the late summer and early autumn there are 
not even thunderstorm rains to break the drought. The 
general warming spreads even to the equator, though the 
rise of temperature is less there than in other parts of the 
world, and the temperature difference between low and high 
latitudes is greatly diminished. 

The Greenland ice-sheet is now retreating very rapidly, 
and soon breaks up into a number of isolated masses of dead 
ice in the valleys. The warming up of the oceans breaks up 
the ring of pack ice surrounding the Antarctic ice-mass, and 
that, too, has withdrawn within the limits of the continent 
and is now in full retreat. All this ice-water added to the 
sea raises its level still further, and helps the influence of the 
warm oceans to penetrate into the land. As the mountains 
are worn down they offer less hindrance to the winds, and there 
is less forced ascent of air and less orographic cloud and rain. 
Cyclones are now fewer (though destructive hurricanes still 
occur in the tropics and sometimes travel into temperate 
latitudes), and altogether the general rainfall is decreasing. 
Since vertical motion is the most effective agent in causing 
the air to condense its moisture, and the diiest air is that which 
has parted with its water vapour by being raised, and then 
has again descended to the surface of the earth, this general 
decrease of vertical motion gives rise to a general increase 
in the relative humidity, so that evaporation decreases, and 
the surface of the oceans does 'not lose so much heat in this 
way. The wind velocity also is generally smaller, and this 
again decreases the evaporation. When all the ice has 
finally disappeared, the surface of the oceans is everywhere 
warm, and owing to the prevailing poleward component of 
the surface winds, this mass of warm water is almost everywhere 
moving from lower to higher latitudes, carrying genial con- 
ditions into the neighbourhood of the poles. At first the 
depths are still filled with cold water a relic of the ice-age 
but this cold mass receives no fresh supplies, and is gradually 
warmed up by earth-heat and by conduction from the surface. 
The earth has now entered on the " non-glacial " stage, and 


conditions will remain sensibly unaltered for millions of years, 
until a fresh manifestation of the internal forces of the earth 
raises new mountain ranges to begin the cycle afresh. The 
climatic conditions at the height of a " non-glacial " or warm 
period are of such great interest that a fuller description of 
them is reserved for the next chapter. 

For a fuller description of the cycle of compression, mountain- 
building and erosion, the reader is referred to a book by 
J. H. F. Umbgrove (8). 


(1) WAGNER, A. " Klimaanderungen und Klimaschwankungen." Braun- 

schweig, Die Wissensch.y Bd. 92, 1940. 

(2) JEFFREYS, H. " The earth, its origin, history and physical constitution.*' 

2nd ed., Cambridge, 1929. 

(3) MANLEY, G. " Glaciers and climatic change, some recent contributions." 

London, Q. J. R. Meteor. Soc., 72, 1946, p. 251. 

(4) LE CONTE, J. " The Ozarkian and its significance." J. GeoL, Chicago, 

7, 1890, p. 525. 

(5) RAMSAY, W. " Orogenesis und Klima." Ofversigt af Finska Vetenskaps Soc. 

Forh., 52, 1910. 

. " The probable solution of the climate problem in geology." Geol. 

Mag., London, 61, 1924, p. 152, and Washington, Smithsonian Rep., 1924, 

P- 237- 

(6) ANTEVS, E. " The last glaciation." New York, Amer. Geogr. Soc., Research 

Ser., no. 17, 1928. 

(7) MARTONNE, E. DE. " Trait de geographic physique." 2nd ed., Paris, 1913 

(8) UMBGROVE, J. H. F. " The pulse of the earth." 2nd ed., The Hague, 1947 



WE have seen that during long periods of geological 
time the earth seems to have been free from ice, so 
that even if we accept the theory of Continental 
Drift (set out in Chapter XIII.) it is still necessary to believe 
that the climatic zones were much less marked than at present. 
These periods include a large part of the Palaeozoic, almost 
all the Mesozoic, and about half of the Tertiary. Remem- 
bering that they were also periods of low relief and very slow 
denudation, we must suppose that the time intervals represented 
by the rocks of these periods were very much greater than the 
time intervals represented by the rocks of the more active 
periods, and that this type of climate has, in fact, prevailed 
during the greater part of geological time. Its most striking 
feature is the appearance of vegetation of sub-tropical or warm 
temperate aspect in very high latitudes. These periods have 
been termed " pliothermal," though I prefer the simpler word 
" warm," which means the same. 

In the preceding chapters the various meteorological features 
of the warm periods have been deduced, and it is only necessary 
here to collect these results and so present a more or less 
connected picture of their climate and weather. 

Let us imagine that we are voyagers in one of these favoured 
periods of antiquity, sailing northward from the equator on 
a voyage of discovery towards the pole. We have set out 
from a low marshy western shore, through which a broad 
but shallow river moves sluggishly towards the sea, winding 
in endless curves over a vast plain which stretches as far as the 
eye can reach. Under the influence of a light easterly breeze 
we sail slowly towards the north-west. The sky is half covered 
by woolly cumulus clouds, which now and again thicken and 
darken to a passing shower, with perhaps a burst or two of 
thunder and a slight squall. The air is warm and moist, 
but not unpleasantly so, though we are conscious of a feeling 



of lassitude which makes us disinclined for effort, either 
physical or mental. As we pass northward, the barometer 
rises, the wind backs to north-east, the sky becomes ever 
clearer, and the air more bracing. Detached cloudlets are 
now the rule. Showers still fall at long intervals, but mostly 
in the early hours of the night. On one or two islands that 
we pass we find evidence in the fallen trees that occasional 
hurricanes occur, but they are very rare, and we are not 
greatly troubled by the chances of meeting one of them. 
The sea is very warm and blue, and is teeming with life ; 
the nights are almost as warm as the days. Still farther 
north, and we enter a region of light variable winds and 
calms. At times we are becalmed for several days, while 
overhead the sun blazes through a cloudless sky. Then 
there is a breath of wind from the north-east or from the 
south-east, and a few clouds gather, with perhaps a shower 
of rain. We are now in the western half of the ocean, and 
all the time we are drifting slowly northward on a great 
ocean current. The barometer falls very gradually, the 
southerly winds become more frequent and stronger, they 
draw round to south-west, and, finally, with all sails set, we 
bear away to the north-east. The weather, however, does 
not differ appreciably from that experienced farther south ; 
it is still very warm and for the most part sunny, with a few 
scattered clouds. The only difference is that now and again, 
perhaps once a month, the barometer drops a few tenths of 
an inch, the southerly wind freshens, and a uniform but not 
heavy cloud canopy covers the sky, while for a few hours it 
rains almost steadily. Then the wind veers to west or north- 
west, the sky clears, and after a few showers the steady fine 
weather sets in again. It is noticeable, however, that even 
when the wind blows straight out of the north it is never cold, 
although the season is only late winter. 

For some time now we have been sailing northward, with 
a low palm-fringed coast in view to the eastward. Beyond 
the belt of palms, which marks the limit of the occasional 
thunder rains brought by the daily sea breezes, we catch 
glimpses of a series of low rounded sandy hills, which seem 
to be almost devoid of vegetation. Here and there a line 
of reeds marks the course of a stream bed. Mostly, they are 
dry and withered, and the channel opens out on to a glittering 



level of white, where a layer of salt or gypsum encrusts thedried- 
up floor of a temporary pool, but occasionally the reeds are 
fresh and green, marking either a more permanent river, or 
the channel taken by the waters of a recent storm. Only 
once, however, do we see it actually raining over the land. 
On that occasion, the low rounded hills which everywhere 
form the background of the landscape, and which usually 
stand out clearly against a sky of intense blue, seem to support 
a mighty column of cloud, in which we can see the play of 
lightning flashes and the deluging rain. We know that 
some of the dry channels will soon be rushing torrents of 
water, and that the apparently lifeless hollows of the plain 
will wake to swarming life. One of the fishes the Ceratodus 
or lung-fish is specially adapted to the chances of this life, 
since it can breathe air or water at will, but how the other 
animals survive the periods of desiccation is a mystery. 

As we pass northward, the desert character of the land 
becomes less marked, and in about the latitude of London 
we decide to land and carry out some explorations on foot. 
It is now early in March, but the weather has almost the 
character of a fine English June. The vegetation is very fresh 
and green. As we go inland across Europe the air becomes 
cooler and crisp though not cold. Now and again, after 
a few hotter days, there follows a heavy thunderstorm, and this 
is the only rain that falls at this season. Later in the year 
all this expanse of country will be burnt brown in the steady 
heat of a rainless summer ; then as autumn draws on, the 
sun will lose some of its power, and the year will draw to its 
close in a spell of perfect weather, such as even now we 
occasionally meet with in September. Just about midwinter 
perhaps one or two mild storms from the ocean to the westward 
will pass across the land. 

Here we must interrupt our voyage for a moment to explain 
that the narrative would differ slightly according to whether 
we were in the Mesozoic or in the Early Tertiary world. A 
vivid picture of Jurassic Oxfordshire has been drawn by 
W. J. Arkell (i). In a clear shallow sea grew a complex of 
true coral reefs intersected by narrow channels. The reefs, 
which are nowhere very thick, are interspersed with beds 
of limestone resulting from their erosion, and it is inferred 
that the corals grew on a rising sea floor. The corals are 


obviously found where they grew, and it is very improbable 
that the sea temperature was less than 60 F., more than 
5 F. warmer than the present temperature of the Atlantic 
west of Ireland, which itself has the highest sea temperature 
for its latitude anywhere in the world. In the Early Mesozoic 
even frost seems to have been unknown, and in Europe the 
rainfall was not heavy enough to balance the evaporation. 
The prevailing continental deposit was a red desert sandstone, 
and in our own country we have in Leicestershire a true 
" fossil desert," where wind- worn pillars of granite stand up 
into the Triassic sands which ultimately buried all but their 
highest summits. A very good picture of the conditions in 
the type of country which prevailed over a large part of the 
Mesozoic world is conjured up by the finds of dinosaur 
eggs in Mongolia. The term " desert " is probably not 
strictly applicable, for there must have been enough vegetation 
to support these great reptiles. Probably, however, the 
vegetation was limited to the larger water-courses which 
received the drainage from a considerable area, and the 
intervening country was a sandy waste, in which the dinosaurs 
buried their eggs to be hatched by the heat of the sun. 

Still earlier, in the Permian, a large part of Europe had 
been occupied by a great salt inland sea with desert shores, 
a European Caspian. This lake was saturated with salt, and 
a striking witness to the stability of the climate is the fact that 
year by year for ten thousand years the excess of salt was 
deposited in regular layers. In summer the deposition was 
mainly in the form of gypsum, but this mineral is much more 
soluble in cold than in warm water, and as the sea cooled in 
winter the deposit of gypsum ceased and rock salt took its place. 
From the evidence of similar evaporating solutions in the 
Sahara, Kubierschky, quoted by Koppen and Wegener 
(Chapter XI II.), estimates that the temperature of the lake 
varied from about 60 F. in winter to as much as 95 F. in 
summer. In addition to the main mass of salts there are some 
isolated layers, apparently formed when adjacent hollows 
were flooded at high lake stages, the water afterwards 
evaporating and the deposited salts being covered by desert 

In the Early Tertiary, after the brief cold spell which 
ushered in the Eocene had passed, there was another long 


period of warm climate, though not nearly so hot and dry 
as the Early Mesozoic. In fact, although summer was probably 
dry, the spring appears to have been decidedly wet in 
Europe north of the Mediterranean, and beds of brown coal 
were formed in many localities. The rainfall was still mainly 
of the instability type, however, falling in violent thunderstorms 
accompanied by heavy rain or hail, which stripped leaves 
and twigs from the trees and washed them into the lakes. 
Many of the trees were evergreens, and the deciduous trees 
were in leaf by the end of March, as shown by the relation 
of the leafing to the flowering period. Early in the Miocene, 
however, the leaves of some of the beech trees show signs of 
the action of frost. H. von Ihering (2) estimates the mean 
temperature of central Europe as about 68 F. in the earliest 
Eocene, 74 F. in the main part of the Eocene and Oligocene, 
72 F. in the Miocene and 60-55 F. in the Pliocene. 

In western U.S.A., according to various papers summarised 
by R. W. Chaney (unpublished), the climate in the lower part 
of the Upper Eocene was warm and moist, similar to that 
now found on the borders of the tropics. The mean annual 
temperature was about 68 F. and the annual rainfall 70 
inches, rather uniformly distributed through the year. The 
high rainfall was probably due to the neighbourhood of the 
growing Sierra Nevada. These conditions persisted through 
the Lower Oligocene, but in the Upper Oligocene a 
progiessive slow cooling and desiccation set in. 

We must now resume our interrupted voyage towards 
the pole, but this time we will suppose definitely that we 
are in the Upper Eocene period. As we pass towards the 
Arctic Circle, we are still in a great northward-flowing 
warm current, and the vegetation along the shores continues 
to be very rich, but its character gradually becomes more 
temperate. The skies become cloudier, and steady cyclonic 
rain replaces more and more the thunder rains of Central 
Europe. By the time we have passed the latitude of 70 N., 
we find that misty rainy weather forms the rule and fine 
sunny weather the exception, while the dense forests along the 
eastern shore are frequently hidden by fog. The wind is 
now mainly from the west, and the western shores of the 
ocean, and still more the country some distance inland, begin 
to present a bleak appearance. In winter, the low hills are 


probably snow covered, but there are no glaciers. So we 
come to the pole, in a great open basin filled with warm sea 
water from the south, which circulates slowly round and round 
until it cools and sinks to the bottom. There is no great mass 
of Palaeocrystic ice such as we find to-day, no icebergs even, 
and although the great rivers sometimes bring down a few 
fragments of drift ice in the winter, these soon melt. If we 
had made this journey in the Jurassic period we should have 
seen no ice at all ; instead, we should have found coral reefs 
almost to the Arctic Circle and isolated corals even farther 

This is perhaps a somewhat fanciful picture of conditions 
during the great warm periods, but it is based almost entirely 
on geological evidence, or on logical meteorological deductions 
from the slight differences of temperature which we know to 
have prevailed between different latitudes during those periods. 


(1) ARKELL, W. J. " On the nature, origin and climatic significance of the 

coral reefs in the vicinity of Oxford." London, Q. J. Geol, Soc., 41, 

!935 P- 77- 

(2) IHERING, H. v. " Das Klima der Tertiarzeit." Zj. Geophys., Leipzig, 3, 

7> P- 3 6 5- 





WHATEVER view we take of the major cause of the 
great climatic fluctuations of geological time, there 
can be no doubt that the geographic;;:! conditions 
have always played an important part in at least the local 
distribution of climate. By geographical conditions, we imply 
not merely the bare distribution of land and sea, but also other 
variables, such as the height of the land, the presence of 
important mountain chains, the vegetative % > jvering, the 
movements of the sea in ocean currents, and tvf. existence of 
volcanoes. Consequently, before we can pass lo a discussion 
of the meteorology of the different geological ages, we must 
consider to what extent these geographical factors have varied 
in the past. 

Palaeogeography, or the reconstruction of former geo- 
graphical conditions, is a very difficult science. If we find 
a marine deposit, we know that that particular region must 
have been at sea at the time ; similarly, the presence of a 
deposit obviously laid down on l^.i or in fresh water indicates 
the presence of land, but unt ^ivocal evidence of this kind 
is the exception, especially for the earlier periods, the deposits 
of which have largely been destroyed during the course of 
geological history, or buried so deeply as to be inaccessible. 
We are almost completely ignorant of the sequence of deposits 
on the floors of the oceans. Hence most of the reconstruction 
must be based on inference from a variety of facts. A break 
in a series of marine deposits, otherwise undisturbed, indicates 
that for part of the time, at least, the region was above the 
sea. The gradual change of a bed of marine deposits from 
fine clays to coarse sands as we follow them from one region 
to another, points to the existence of land not far beyond the 
latter. The discovery of the same marine fauna in two 
different localities indicates a sea connexion between them, 
while the presence of different marine faunas of the same age 


suggests a land barrier. Similarly, the presence of similar 
land faunas or floras in distant regions points to a land con- 
nexion, while the presence of different land faunas or floras 
close together points to a barrier which may be a sea channel, 
but may equally be a range of mountains or a desert. It is 
by the gradual collection of such diverse facts as these that the 
science of palacogeography has grown up. 

The earlier geographical reconstructions presupposed that 
the various deposits were laid down practically where they are 
found to-day. A few minor exceptions were recognised ; 
for instance, the intense crumpling of the rocks in the Alps 
and Himalayas shows that the lands on either side of these 
chains were formerly at a greater distance from each other, 
and their approach has forced the intervening rocks to lie 
over one another in great heaps, but these shiftings were 
matters of only a few hundred miles. Some geophysicists, 
notably A. Wegener, have not accepted this limitation, but 
consider that continents have drifted like floating islands over 
the face of the earth, and that the positions of the poles have 
changed greatly during geological time. Discussion of this 
theory is postponed to the next chapter ; here it will be 
assumed that the various deposits were formed where they are 
now found, and that ihe positions of the poles relatively to the 
continents have remained practically unchanged throughout 
geological time. 

We saw in Chapter Vl^ "-hat the mean temperature of 
any latitude at the present c'ay depends to a considerable 
extent on the percentage of the area, along a belt 20 wide 
centred on that latitude, which is occupied by land. Near 
the equator the effect of a large land-mass is to raise the 
temperature slightly, but in high latitudes land lowers the 
mean temperature, and especially the winter temperature, 
much more effectively. For reasons to be given later, it is 
necessary to limit the discussion to the mean temperature of 
the regions between the poles and latitude 40. Hence the 
area of land north of 40 N. and south of 40 S. is an important 
climatic factor. Owing to the large area of the oceans south 
of 40 S., we are almost entirely ignorant of the land and sea 
distribution in that part of the world during geological times, 
so that this variable becomes in effect the area of land north 
of 40 N. This area has been measured from the composite 


charts given by Th. Arldt (i). Numerous reconstructions 
of land and sea distribution in different periods have been 
published by various authors, which naturally differ widely, 
and Arldt has combined these, indicating the areas shown as 
land by all authors and also those shown as land by some 
authors and as sea by others. On examining the measure- 
ments, it was found that in the different geological periods the 
areas shown as land by all authors were on the average about 
the same as the present area of land. Since we have no 
reason to suppose that during the Miocene period, for instance, 
the land area in the northern temperate and polar regions 
was much greater than at present, the conservative measure 
given by the areas shown as land by all authors seems the best 
measure to adopt. 

The effect of land areas on temperature increases very 
rapidly as we approach the poles. This was allowed for 
by weighting the land areas in different latitudes according 
to a scale derived from the investigation of " Continentality 
and Temperature" (Chapter VIII.). Giving a unit area 
of land in latitude 60 a weight of i, the scale adopted was 

80, 3 ; 70, 2 ; 60, i ; 50, 0-6 ; 40, 0-2. 

The figure which would be obtained if the whole hemi- 
sphere north of 40 N. were occupied by land was given the 
value 100, so that the figures under (i) of Table 14 are 
percentages of a full land covering. 

The second variable is the average height of the land. 
There is a consensus of opinion among geologists that this 
has varied extensively during geological history. Thus 
T. C. Chamberlin (2) writes : 

" It is generally agreed that the present altitude of the 
continents is greater than their mean elevation during geologic 
history. Geologists recognise at least two stages in which 
the continents were exceptionally high and broad ; that 
which attended the transition from the Palaeozoic to the 
Mesozoic Era, and that which attended the transition from 
the Tertiary to the present epoch. The existing stage thus 
falls in one of the most notable stages when continental 
elevation and breadth were greatest, though perhaps not at 
its climax. The latest estimate of the present mean elevation 
of the land gives 2,500 feet. The mean elevation of the great 




Upper Proterozoic 

Eocambrian . . 
Late Cambrian . 
Ordovician . . 
Silurian .... 

(') (a) (3) 

Conti- Eleva- Ocean 

nentality. tion Currents. 

Per (unit Per 

cent. 100 feet.) cent. 

Lower Devonian 
Middle Devonian 
Upper Devonian . . 

Lower Carboniferous 
Middle Carboniferous 
Upper Carboniferous 

Lower Permian 
Middle Permian . 
Upper Permian . . 

Lower Trias . . . 
Middle Trias . . . 
Upper Trias . . . 

Rhaetic .... 


Middle Jurassic . 
Upper Jurassic . . 

Lower Cretaceous 
Middle Cretaceous 
Upper Cretaceous . 

Lower Eocene . . 
Upper Eocene . . 
Oligocene .... 
Miocene .... 
Pliocene .... 
Pleistocene .... 











5 1 










Recent 50 











J 7 



4 6 


4 6 


































Mean Tempera- 
ture, 4O-90 N. 

"Ob- " Calcu- 
served." lated." 

29 F. 40 F. 

5 1 








4 J 



















Table 14. Values of geographical elements and of mean 

peneplains is a matter of judgment rather than of knowledge, 
but no one would probably put the elevation at much more 
than a third of this. Probably a third is too high." 

Similarly, H. Jeffreys (3) in discussing the discrepancies 
between the age of the earth calculated from the rate of 
accumulation of sedimentary rocks or of salt in the ocean 


and the much greater age calculated from the data of radio- 
activity, points out that the former " amount to a proof that 
the present rate of denudation is several times greater than the 
average of the past." 

It has been pointed out in Chapter X. that the earth's 
surface has passed through a series of cycles, each cycle 
consisting of a relatively short stage of intense mountain- 
building, in which the rocks were thrown into great folds 
and ridges and the average elevation of the land above the 
sea became very great, followed by a long stage of quiet 
conditions, in which the forces of denudation lowered the 
level of the land, rapidly at first, and then more and more 
slowly. We are at present living shortly after one of the 
periods of mountain-building and elevation, and the average 
level of the land is consequently high, though not so high as 
it was during the Quaternary period. At the close of one of the 
long quiet periods the average level must have been very much 
less than it is now, and was probably only a few hundred feet. 
At such times, conglomerates and coarse marine sandstones 
are almost entirely absent, and the bulk of the sedimentary 
rocks is composed of limestones and very fine clays or shales. 

Since we cannot hope to know the average height of the 
land in the different geological periods directly, we have to 
take as a measure the amount of disturbance of the rocks 
caused by mountain-building during that period. The 
column under (2) in Table 14 and the curve of height in Fig. 
24 are based on a diagram given by E. Dacque (4, p. 449). 
Dacque places the chief periods of mountain-building in the 
Algonkian or Late Proterozoic, the Late Carboniferous, and 
the close of the Tertiary, with minor periods at the close of 
the Silurian, in the Cretaceous, and in the Early Tertiary. 
The longest orogenetically quiet period appears to have fallen 
in the interval from Middle Permian to Middle Jurassic. 
There are, however, two points in which Dacque's curve 
seems to require modification. Arldt points out (i, p. 711) 
that the "alpine" character of a mountain chain a large 
number of separate peaks of approximately the same height - 
occurs only as a result of glaciation ; before glaciation, the 
system, though it may be of considerable height, possesses 
only rounded contours of the foot-hill type. The conversion 
of the rounded contours into the broken alpine type, and the 



9 -d 


removal of the resulting detritus, represents a great decrease 
in the load on the underlying plastic layer of the earth, and 
consequently leads to a further elevation. Hence it seems 
probable that the greatest height of the mountains occurred 
after the main periods of mountain-building. Of course 
this does not mean an increase in the average height of the 
earth's surface, but for the meteorological processes involving 
forced ascent of air, a broken " alpine " surface is as effective 
as a rounded surface of much greater area. The second point 
is concerned with the depression of the sea-level during glacial 
periods owing to the accumulation of water in the great ice- 
sheets, which may add several hundred feet to the effective 
height of the land above sea level. For these reasons, and also 
because Dacque's curve represents mountain-building activity, 
while what we require is the effective average height of the 
land, which naturally lags somewhat behind the process of 
uplift, I have modified Dacque's curve slightly, especially in 
the Quaternary period. The height of this modified curve at 
the end of the Quaternary was 19 (the unit being one-thirtieth 
of an inch on Dacque's diagram), and this figure represents 
the present height of 2,500 feet. The lowest value on the 
scale, i, was considered to represent a mean elevation of 500 
feet, and the average height of the land during the Quaternary, 
44 on the scale, was taken as 3,500 feet. Through these 
three points (i, 500; 19, 2,500; 44, 3,500) a smooth curve 
was drawn, and this curve was employed to convert the scale 
elevations into estimated mean heights in hundreds of feet. 
Of course these figures have no pretension to any great degree 
of accuracy ; the conversion was undertaken merely because 
Dacque's curve of mountain-building, if taken directly as a 
curve of height, seemed to exaggerate the heights of the 
disturbed periods relatively to the present far too much. 

The third geographical variable which we have to consider 
is the oceanic circulation. At present the Arctic basin is 
nearly surrounded by a land-ring, which is effectively broken 
only by the Atlantic gap, the Bering Strait being too narrow 
and shallow to admit an appreciable current. Even the 
whole of the Atlantic gap is not occupied by the warm current, 
its western side being occupied by the cold ice-bearing East 
Greenland Current, but this condition is probably only present 
when the Arctic Ocean is ice covered. During the warm 


periods, the surface cold currents were probably very limited 
in area ; the gaps in the circum-polar land-rings were probably 
occupied almost entirely by warm currents, directed towards 
the poles, strongest near the eastern sides of the gaps and 
diminishing in strength towards the west. Near the western 
shores of the oceans there might be local cold currents of slight 
intensity issuing from rivers or narrow channels between 
islands, which would be able to maintain their identity for 
a time owing to their slight salinity. In order to obtain 
comparable measures of the amount of heat carried to the 
north polar lands during the different geological periods, I 
measured the width at 60 N. of those gaps in the circum-polar 
land-mass which were directly connected with tropical or 
sub-tropical seas. For each gap double weight was assigned 
to the first 20 of breadth, single weight to the second 20, 
and half weight to breadths beyond 40. The results are 
shown in Table 14 and in the curve labelled " ocean currents " 
in Fig. 24. The figures are expressed in percentages of the 
oceanic effect which would be produced if the only land 
consisted of five long narrow islands directed from south to 

After the publication of the first edition a series of 
experimental reconstructions of ocean currents were described 
by P. Lasareff (see p. 78). These bear out the general 
lines of the estimated effect of ocean currents given in Table 14. 

The fourth factor which has been considered as " geo- 
graphical " is the amount of volcanic action. Numerous 
climatic roles have been assigned to volcanoes by different 
investigators, some of which are favourable and others 
unfavourable to high temperatures. The cooling effect of 
volcanic dust postulated by W. J. Humphreys (Chapter VI.) 
seems to be the best founded of these roles. As Humphreys 
points out, the amount of volcanic dust discharged into the 
upper air depends on the explosive eruptions and not on the 
total amount of volcanic action, but it does not seem possible 
to obtain a curve of explosive volcanic activity only. Con- 
sequently, we have to be content with a measure of the total 
volcanic activity in each period, based on the thickness of 
volcanic rocks, especially lavas. For this purpose the 
admirable summary of volcanic activity given by Arldt (i) 
was converted into figures on a comparative scale of o-io. 


Finally, values were assigned for the mean temperature, 
based on a curve given by Dacqu to show the zonal 
differentiation of climate. While it is not difficult to accept 
Dacque's curve as having some value for the Mesozoic and 
Tertiary periods, we get into difficulties as soon as we go back 
to the Palaeozoic. The Upper Carboniferous especially, 
with its apparently tropical forests in temperate latitudes 
accompanied by an enormous glaciation near the present 
equator, is a meteorological paradox. For the purposes of 
comparison, figures were assigned, but they are very doubtful. 
In considering Dacque's curve, it was assumed that the mean 
temperature of the equatorial regions (apart from the Permo- 
Carboniferous period) had remained constant, and that the 
curve, therefore, gave the variations of temperature in the 
temperate and polar latitudes in the Northern Hemisphere ; 
to be precise, in the area between 40 N. latitude and the 
North Pole. This cuts out the areas which were most heavily 
glaciated during the Upper Carboniferous, and the tempera- 
tures to be assigned to that period were therefore not so low 
as those of the Upper Proterozoic and Quaternary periods, 
although the total amount of land ice present during the 
Upper Carboniferous was probably greater than the amount 
at any other stage of the earth's history. On the other hand, 
the Upper Carboniferous presents evidence of a considerable 
amount of glaciation even in North temperate latitudes, a 
point which is discussed further in Chapter XV. Moreover, 
the faunal changes at this time, and especially the great 
extinction of corals, indicate a great lowering of the temperature 
of the sea. It was therefore assumed that there was a con- 
siderable fall of temperature during the Upper Carboniferous 
even in North temperate and polar latitudes, though not so 
much as in the Upper Proterozoic or the Quaternary. 

In order to obtain numerical measures it was necessary to 
find some means of calibrating Dacque's curve. The mean 
temperature of the area between 40 N. and the North Pole 
is at present 33 F. For the Middle Jurassic, the January 
temperature of the north polar basin calculated in Chapter 
VIII. was 44-5 F. and the annual range was 13*5 F., giving 
a mean annual temperature of 5 1 F. 

The variation with latitude over the oceans was small ; 
on the other hand, the winter temperatures over the interior 



of the continents must have been several degrees lower than 
those near the oceans. As a rough approximation, a value 
20 F. above the present, or 53 F., was accepted as the mean 
temperature of the whole region in Middle Jurassic times. 
This value seems reasonable from a consideration of the 
biological evidence. 

The mean temperature in the Pleistocene period was taken 
as 5 F. below the present mean. The maximum decrease 
in the mean annual temperature calculated from the lowering 
of the snow-line was more than 20 F. in Scandinavia, 20 F. 
in East Anglia, 11 F. in the Alps, and 7 F. in Japan ; on 
the other hand, the Pacific Ocean was little affected, and it is 
not improbable that over the interior of Asia the winter 
temperatures were higher than now. When the interglacial 
periods are taken into account, a mean decrease of 5 F. over 
the whole area north of 40 N. seems to be a reasonable 
estimate. It happened that the differences of 5 F. between 
the Pleistocene and the Present, and 20 F. between the 
Present and the Middle Jurassic, were actually proportional 
to the differences measured on Dacque's curve, and while 
this is probably nothing more than a coincidence, it greatly 
facilitated the conversion of the scale of this curve into a 
temperature scale. 

The next step was to determine how far the various elements 
continentality, elevation, ocean currents, and volcanic 
activity were responsible for the mean temperature. For 
this purpose the figures in the column of Table 14 headed 
" Mean Temperature, 4o-go N., Observed," were 
" correlated " with the figures in the columns headed 
" Continentality," " Elevation," " Ocean Currents," and 
" Volcanic Action." The figures were divided into two 
groups, the doubtful figures for the Upper Proterozoic and 
Palaeozoic being separated from the much more reliable 
figures for the Mesozoic, Tertiary and Recent. The cor- 
relation coefficients are given in Table 15. 

Temperature with 
Continentality. Elevation. Ocean Currents. Volcanoes. 

Palaeozoic .... 52 66 +*37 --50 

Mesozoic to Recent 37 -72 +'51 "ii 

Table 15. Correlation coefficients between temperature and 
geographical conditions. 


A correlation coefficient of + 1 indicates that the fluctuations 
of the two variables considered are exactly proportional ; 
a coefficient of i indicates that the relationship is exactly 

These coefficients agree with our expectations in showing 
that extensive land areas, a high level, and extensive volcanic 
eruptions are all associated with low temperatures, while 
open connexions between the Arctic Ocean and equatorial 
seas are associated with generally high temperatures. The 
chief difference between the two periods Pakeozoic and 
Mesozoic to Recent, lies in the importance of volcanic 
eruptions, which appear to have been much more effective 
in the former than in the latter. This is largely due to the 
very great volcanic activity which prevailed during the 
Permo-Carboniferous glacial period, which dominates the 
first half of the climatic curve. 

Correlation coefficients show how closely two variables 
are connected, but they do not give immediately the 
quantitative effect which one variable has on the other. 
This is given by the " regression coefficient/ 5 which is the 
average amount of change in one variable associated with 
a change of one unit in the other variable. For instance, it 
was found that the regression coefficient of mean temperature 
(Fahrenheit degrees) in terms of continentality (per cent.) 
during the Palaeozoic period was 0-31. This means that 
an increase in the continentality by one per cent, is associated 
with a decrease in the mean temperature north of 40 N. by 
o3iF. The regression coefficients calculated from the 
correlation coefficients in Table 15 are shown in Table 16. 

Factor .... Continentality. Elevation. Ocean Currents. Vulcanicity. 

Unit .... i per cent. 100 feet i per cent, i (scale o-io) 
Change of Temperature 

Palaeozoic .... 0-31 0-38 +0-28 1-68 

Mesozoic to Recent 0-32 0*47 +0*28 0*42 
" Theoretical " (see 

below) .... -0-35 -0-3 (+0-3) -0-5 

Table 16. Effect of a change of one unit on the mean 
temperature in F. 

With the exception of the figures for vulcanicity, there 
is remarkably good agreement between the two periods. 


Let us now consider the factors separately. An increase 
of one per cent, in the land area north of 40 N. is found 
to decrease the mean temperature by 0-31 F. In discussing 
the effect of continentality on temperature at the present 
day (Chapter VIII.), I obtained expressions for the effect of 
land in different latitudes on the mean temperature in January 
and July. The effect of an increase of the land area in any 
region is made up of two parts, a general decrease in the mean 
temperature over the whole belt of latitude, and an additional 
local decrease in the neighbourhood of the new land area. 
The average effect of an increase of one per cent, in the land 
area, calculated from the present distribution of temperature 
in relation to the distribution of land and sea, is a decrease 
of 0-22 F. in the " zonal" temperature (p. 150), and the local 
effect, if spread out over the whole zone, would be equivalent 
to an additional decrease of o- 13 F., making a total lowering 
of temperature by o35 F. This is the " theoretical " figure 
of Table 16 ; it is in good agreement with the figures 0-31 F. 
and 0-32 F. obtained from the regression equations. 

The effect of elevation on temperature at present is well 
known. It is very close to an average decrease of 0-5 G. 
per 100 metres or 0-3 F. per 100 feet of elevation, which is 
given as the " theoretical " figure in Table 16. 

The effect of ocean currents is complicated by the great 
cooling power of floating ice described in Chapter I. For 
our representation of the warm periods, we may take the 
arithmetical mean of the temperatures over the Arctic Ocean 
calculated in Chapter VIII. for the Upper Jurassic and 
Middle Eocene periods, viz., 43 F. Thus we have the 
following data : 

Warm Period. Present. 

Ocean Currents. Arctic Temperature. Ocean Currents. Arctic Temperature. 

Glacial. Non-Glacial. 

44 per cent. 43 F. 17 per cent. 18 F. 24 F. 

If we take the glacial temperature at present, we have a 
difference of 61 F. corresponding with a difference of 27 per 
cent, in the ocean currents ; if we take the non-glacial, we 
have a difference of 19 F. Now, in geological time, non- 
glacial conditions in the Arctic Ocean have been the rule 
and glacial conditions the exception. If we take the ratio 


of occurrence of the two conditions as five to one, and combine 
the two figures 19 F. and 61 F. in that proportion, we obtain 
a weighted mean of 26 F. for a difference of 27 in the ocean 
currents. This figure refers to winter over the Arctic Ocean ; 
the difference between summer and winter is probably not 
great, but the effect diminishes rapidly southward, and is also 
less over the land than over the oceans. If the average effect 
over the whole area north of 40 N. is one-third of that over 
the Arctic Ocean itself, or 9 F., we obtain an increase of 
temperature of about 0-3 F. for an increase of one per cent, 
in the effect of the ocean currents. The amount given in 
Table 1 6 is 0-28 F. 

The effect of volcanic action is difficult to discuss because 
of the arbitrary nature of our scale of o-io. W. J. Humphreys 
(Chapter VI.) considers that during the past 160 years the 
mean temperature of the earth has been lowered i F. by 
volcanic dust. If the value of 2 for the present vulcanicity 
is correctly assigned (a very large "if"), this is equivalent 
to a decrease of temperature by 0-5 F. for an increase of 
one in the scale of vulcanicity. The corresponding value 
found for the Mesozoic to Recent periods is 0-42 F., which 
is a good agreement. On the other hand, the figure for the 
Palaeozoic, 1-68 F., is very much greater, and suggests that 
the amount of volcanic dust present during the Upper 
Proterozoic and Upper Carboniferous glaciations was much 
greater than is indicated by the values of 9 and 10 on the 
linear scale. The figure for the Upper Carboniferous should 
probably be nearer 40 than 10. 

From the correlation coefficients given in Table 15, we 
see that the variations of climate during geological time 
have been associated to some extent with the variations 
of all four of these factors continentality, elevation, ocean 
currents, and volcanic action. But the curves in Fig. 24 show 
also that these factors of climate have a close relationship 
among themselves. When the continents were generally 
lofty they were also extensive, and the passages between 
them along which ocean currents could penetrate into high 
latitudes were few and narrow, while volcanic action was 
greatest in periods of mountain-building. Hence part of 
the effect of great continentality in lowering temperature 
may be due to the great elevation, weak ocean currents, and 


great vulcanicity which accompany it. In order to determine 
the effect which would follow a change in one factor only, 
while the others remained constant, it is necessary to calculate 
" partial " correlation coefficients. This was done, and the 
results are shown in Table 17. 

Temperature with 
Continentality. Elevation. Ocean Currents. Volcanoes. 

Palaeozoic .... 25 -22 -22 ~ *53 

Mesozoic to Recent -08 -82 +'63 -15 

Table 17. Partial correlation coefficients with temperature. 

In the calculation of partial coefficients, small errors in the 
original data are apt to make a great difference in the final 
result. The figures of continentality and ocean currents 
for the Palaeozoic are uncertain, and the partial coefficients 
for this period have very little value. In particular, the 
negative coefficient between ocean currents and temperature 
is obviously wrong, since one cannot conceive a state of 
affairs in which a wide connexion between the polar and 
equatorial oceans brings about a Jow temperature in temperate 
and polar regions. Probably the only significant figure is the 
high correlation between volcanic activity and temperature. 
For the Mesozoic and later periods our data are more exact 
and the partial coefficients show that the most important 
geographical conditions which determine temperature are 
the average elevation of the land and the volume of the ocean 
currents. From these partial coefficients we obtain the 
following formula for calculating the temperature in any part 
of the Mesozoic or Tertiary : 

Temperature (F.)~48 o 13 X Gontinentality (per cent.) 0-45 
X Elevation (hundreds of feet) +0-43 X Ocean Currents (per 
cent.) o 26 X Vulcanicity (o-io). 

It will be noticed that these coefficients differ somewhat 
from those given in Table 16. The effect of continentality 
appears to be greatly reduced ; this is because the cooling 
power of land is due partly to the mountain systems usually 
found somewhere in a large land-mass, partly to the barriers 
which large land-masses place in the way of ocean currents, 
and only partly to actual cooling by radiation from the surface 
of the land. All three of these effects are included in the 


coefficient in Table 16, but in the equation given above, the 
first two have been eliminated, leaving only the purely local 
radiation effect. It may be only a coincidence, however, that 
the value of this local effect in the equation given above 
(0-13) is exactly the same as the local part of the total 
effect of continentality at present, as described on page 212. 

The theoretical temperatures given by this equation are 
shown in column (6) of Table 14. The calculation was 
extended to the Palaeozoic, although the equation is based 
only on the data since the beginning of the Triassic, because 
we cannot suppose that the physical laws of climate have 
changed, and the equation deduced from the later periods 
agrees with what we know of those laws. The discrepancies 
shown by the Palaeozoic are no doubt due partly to our 
incomplete knowledge of the geographical conditions of this 
era, but probably partly also to errors in the temperatures 
deduced from the records of the rocks. The most noticeable 
feature is that until the Middle Triassic the " calculated " 
curve is generally above the " observed " curve. The three 
great glacial periods of the Upper Protcrozoic, Upper Carbon- 
iferous, and Quaternary stand out clearly ; following each 
one of them the " calculated " curve rises more rapidly than 
the " observed " curve, as if the earth took a long time to 
warm up again after the crisis of the glaciation had passed. 
We have good reason to believe that this is true of the Recent 
period, the temperature being kept lower than it should be by 
the relics of the Quaternary ice-sheets in Antarctica and 
Greenland, and by the low temperature of the great body 
of sea water. The delay in warming up after the other ice- 
ages may be due to similar causes, in which case the statement 
sometimes made, that until the Quaternary the oceans had 
never been generally cooled, is incorrect. There is, in fact, 
a large amount of biological evidence that the oceans became 
cold during the Upper Carboniferous. It is possible, however, 
that the delay in these cases is more apparent than real. It 
is often difficult to determine the exact horizon of a glacial 
deposit, and if a few such deposits are placed too high in the 
series, they will make the stage following the glacial period 
appear colder than it actually was. This explanation may 
apply to the apparently low temperature of the Early 
Cambrian, but scarcely to the discrepancy of the Upper 


Permian and Lower Trias, and on the whole, I believe this 
delay in warming up after an ice-age to be a real phenomenon. 
For the Pliocene, the " calculated " temperature is much 
lower that the " observed." It is a very striking fact, which 
has often been commented on, that the Quaternary glaciation 
did not coincide with the period of greatest elevation, but 
lagged considerably behind it. The cause of this lag was 
discussed on page 179. 

There is a steep drop in the " calculated " curve in the 
Lias (Lower Jurassic) which is barely shown on the " observed " 
curve. The Liassic period, so far as we know, had no glaciers ; 
probably the distribution of mountain ranges in relation to 
moist winds was not suitable. In the absence of ice, the other 
factors of low temperature fail to produce their full effect. 
The drop in the " calculated " curve during the Cretaceous, 
on the other hand, is almost equally marked on the " observed " 
curve ; in this instance we have evidence in the erratics of the 
English chalk that either shore ice or glacier ice occurred 
somewhere in the Northern Hemisphere and that floating 
ice was present on the chalk seas. 

The relations between the " observed " and " calculated " 
temperatures, given in Table 14, since the beginning of the 
Carboniferous have some points of interest. It will be 
noticed that when the " calculated " temperature is below 
39 F. the " observed " temperature tends to be below the 
;c calculated " temperature, the mean values for the five cold 
periods (Upper Carboniferous, Lower Permian, Pliocene, 
Quaternary, Recent) being : " observed," 32-4 F. ; " cal- 
culated," 338F. When the " calculated " temperature 
lies between 39 F. and 50 F., the " observed " temperatures 
tend to be higher than the " calculated," the mean values for 
fourteen moderate periods being: "observed," 47-4 F. ; 
:c calculated," 45-4 F. When the " calculated " temperature 
is above 50 F., the " observed " temperatures are again 
lower than the " calculated," the mean values for the warmest 
periods being : " observed," 50-0 F. ; " calculated," 52-3 F. 
A. decrease of the "calculated" temperature from 52-3 to 
4.5-4 F., or 6 '9 F., is associated with a decrease of the 
"observed" temperature from 50-0 to 47-4^., or only 
2-6 F., a fall of less than 0-4 F. in the " observed " tem- 
perature for a fall of one degree in the " calculated " 


temperature. On the other hand, a decrease of the " cal- 
culated " temperature from 45-4 to 33-8 F., or ii-6F., 
is associated with a decrease of the " observed " temperature 
from 47-4 to 32-4 F., or 15 F., a fall of i'3F. in the 
" observed " temperature for a fall of one degree in the 
" calculated " temperature. This result may be due to one 
of three causes : It may be accidental, due to the chance 
run of the figures, or it may be due to an error in the scale 
of the " observed " temperatures, owing to which the tem- 
peratures of the moderately warm periods are overestimated 
compared with those of both the warm and the cold periods. 
On the other hand, it may represent a real phenomenon, a 
unit change in the geographical factors making less difference 
to the mean temperature of a warm period than to that of a 
cold period. Chapter I., and especially Fig. 2, show very 
strong reasons why such a difference should actually occur. 
So long as the climate remains " non-glacial," the change 
of temperature due to a change in the geographical factors h 
limited to the direct effect of those factors. Land in high 
latitudes has a lower mean temperature than sea, so that an 
increase in the land area lowers the mean temperature some- 
what, but this effect at present is partly due to the winter 
snow-cover. The more free the oceanic communication 
between high and low latitudes, the more heat is carried 
by ocean currents, but the effect is also proportional to the 
difference between the temperature of the warm currents and 
that of the main mass of cooler water in the Arctic Ocean, 
and therefore two ocean currents during a warm period do 
not raise the mean temperature twice as much as one of them 
during a cold period. In Chapter X. we saw that the effect 
of an increase of elevation by 100 feet becomes greater the 
higher the average level of the land ; the change from an 
average elevation of 2,500 feet to one of 3,500 feet is responsible 
for nearly as much cooling as the change from 500 to 2,500 feet. 
When the temperature of the polar regions falls below a certain 
level, the climate becomes " glacial," and the cooling power 
"of ice is added to the direct effect of the geographical factors. 
The relations between the " observed " and " calculated " 
temperatures north of 40 N. may be due to this introduction 
of the cooling power of ice when the mean temperature falls 
below about 45 F, over the whole area, which may imply a 



January mean of 27 F. at the pole. If this explanation is 
correct, the ice present in the oceans in high latitudes, and the 
Greenland ice-sheet, lower the mean temperature north of 
40 N. by nearly 10 F., which is in sufficiently good agree- 
ment with the results of the theoretical investigation in Chapter 
I. The changes of mean temperature which take place 
during the transition from a warm period to an ice-age are 


Calculated Temperatures 
$Q 45 40 35 


30 r 






40 1 

35 I 


Fig. 25.^ The transition from a warm period 
to an ice-age. 

shown diagrammatically in Fig. 25. The mean temperature 
north of 40 N. is initially 52 F., as shown at A. As the 
continents emerge and the ocean currents become weaker, 
the temperature falls slowly, as shown by the full line, until 
the point B is reached, when it is 45 F. At this point the 
Arctic Ocean becomes glacial. The temperature now falls 
much more rapidly along the full line BC. The dotted line 
BC X indicates the " non-glacial " temperature due to the action 


of the geographical factors alone, without the intervention 
of the ice, and the distance between the dotted line and the 
full line shows the additional cooling due to the ice itself. 
The arrow indicates the amount of cooling by ice at the present 
time. The broken line shows the corresponding temperatures 
calculated from the regression equation given above, which 
assumes that the relations are linear throughout. The 
three crosses mark the three points determined from the 
comparison of the " observed " and the " calculated " 
temperatures. The diagram seems to agree well with all 
the results previously obtained ; for instance, it indicates 
that the " non-glacial " temperature is now only about two 
degrees below the critical point, and that a permanent increase 
in the general temperature by more than this amount would 
result in the breaking up of the Arctic ice. The good quantita- 
tive agreement between the effects of the different geographical 
factors calculated from purely geological data and those 
deduced from existing conditions, the coincidences in points 
of detail between the observed and calculated curves of 
temperature in Fig. 24, and the fact that the discrepancies 
between the two curves are what we should expect from the 
combination of " glacial " and " non-glacial " periods in the 
same equation, combine to form a very strong body of 
evidence that throughout the greater part if not eill of 
geological time the major variations of climate have been 
entirely controlled by changes in the geographical factors. 
Since a careful calculation of the effects of known causes 
suffices to explain the facts, it is unnecessary to introduce 
hypothetical causes such as variations of solar radiation or 
continental drift to explain the long-period oscillations of 
climate. The more rapid oscillations within the major 
climatic chapters, such as the succession of glacial and inter- 
glacial periods within the Quaternary, may not be explicable 
by changes in the geographical factors. Zeuner's recon- 
struction of the Quaternary sequence (p. 107) fits in well 
enough with the facts to lend some support to the theory 
that such secondary oscillations are due to astronomical 
causes, but variations of solar radiation also remain a 

The topsy-turvy Permo-Carboniferous period, in which 
the greatest glaciation occurred not far from the equator, 


demands a special investigation, which is given to it in 
Chapter XV. 


(1) ARLDT, TH. " Handbuch der Palaeogeographie." 2 vols. Leipzig, 1919. 

(2) CHAMBERLIN, T. C., and Others. " The age of the earth." Philadelphia, 

Proc. Arner. Phil. Soc., 61, 1922, and Washington, Ann. Rep. Srnithson. hist., 
1922, p. 246. 

(3) JEFFREYS, H. " The earth, its origin, history, and physical constitution." 

Cambridge, 1924. 

(4) DACQUE, E. " Grundlagen und Methoden der Palaeogeographie." Jena, 


IN the calculations discussed in the last chapter, the 
assumption was made that deposits were laid down not 
far from where we now find them, or in other words, that 
the positions of the continental massifs relative to each other 
and to the poles have not changed during geological time. 
That assumption has been challenged from time to time, 
but was not seriously countered until A. Wegener ( i ) published 
his well-known theory of continental drift, and supported 
it with a wealth of detail and acute reasoning. For a time 
Wegener's theory was in considerable favour, but it has 
been found to introduce so many difficulties that opinion 
now seems to be that if continental drift ever occurred on the 
scale postulated by Wegener it was long before the beginning 
of the geological record. The final acceptance or rejection 
of Wcgener's theory is a matter for geologists, but inasmuch 
as palaeoclimatological evidence plays a considerable part in 
the working out of the theory, which in turn, if accepted, 
completely alters the aspect of the problem of climatic changes, 
this book would not be complete without a discussion of what 
the theory implies. The theory of continental drift falls into 
two parts, and the truth of one part does not necessarily imply 
the truth of the other part. The first contention is that the 
positions of the continents have changed relative to each 
other during the course of geological time ; that at first there 
was a single large continent (" Pangaea "). This original 
continent split up into various parts which gradually drifted 
asunder, the latest division being the separation of America 
from Europe. The second contention is that there have also 
been radical changes in the positions of these land-masses 
relative to the poles. Of course, if the relative positions of 
two continents change in any way except by means of a 
direct east-west movement, there must necessarily be some 
change of latitude, but the movements postulated by Wegener 
go far beyond this. Regions like Brazil and Central Africa, 


now near the equator, are supposed to have been formerly 
near the South Pole, while other regions now far to the north 
were once close to the equator. 

This power of free movement of the continents depends 
on the difference of constitution between the continental 
massifs and the mass of the earth's crust. The former are 
composed mainly of silicates of alumina, termed Sial for short, 
and are lighter than the rest of the crust, which is mainly 
composed of silicates of magnesia (Simd). The sima is 
continuous and thick, and forms the floor of the deep oceans ; 
the floor of the Atlantic, however, is believed to be covered 
by a thin layer of sial, and this difference between the Atlantic 
and Pacific has given rise to much speculation. The sial 
now consists of a number of separate masses (continents and 
continental shelves) with some smaller detached portions 
(oceanic islands). Under the action of any long continued 
force, however small, the sima acts as a very viscous fluid, 
while the masses of sial are rigid, and may be compared 
to slabs of wood floating in a sea of treacle. This distinction 
between the continental masses and the rest of the crust is 
based on a number of converging lines of evidence, and is 
now generally accepted. 

The argument that the individual continents have been 
formed by the splitting up of an original Pangaea, starts 
with the notable similarity in the shape of the opposite coasts 
of the Atlantic Ocean. Not only is there a great similarity 
of shape, but the structural features on either side of the 
ocean show a considerable degree of resemblance, and through- 
out geological time there has been also a strong likeness 
between the animals and plants. Hence it is supposed that 
America split oft from Europe during the Tertiary period, 
the rift beginning in the south and gradually extending 
northwards. This movement is considered to be still in 
progress, and to have been sufficient to become evident in the 
successive determinations of the longitude of Sabine Island in 
North-east Greenland, which is thus shown to have moved 
westward by nearly a mile in eighty-four years. These 
differences in the longitude, however, as Sir Charles Close 
has pointed out (2), are all within the limits of the probable 
error, and are not sufficient to constitute a proof of the westerly 
drift of Greenland. 


It is true that the distribution of a number of animals 
and plants which are found on either side of the Atlantic 
could be explained much more readily on the hypothesis 
that America and Europe-Africa were in contact not long 
ago, but this involves the supposition of a formerly wider 
Pacific Ocean, against which must be set the distribution of 
a number of animals and plants common to both shores of 
this ocean. 

The second part of the theory is that independent of the 
drift of the continents the earth's axis of rotation is under- 
going a progressive change, which since early Palaeozoic 
times has brought the North Pole from high southern latitudes 
via the Pacific Ocean to its present position. The arguments 
rest entirely on deductions from the distribution of climatic 
zones in past times, and especially on the location of the 
Permo-Carboniferous glaciation. We will return to this 
point in the next chapter. 

The chief weakness of Wegener's theory is the inadequacy 
of the forces which he postulates to move the continents. 
These are twofold a force directed towards the equator and 
a force directed towards the west. The force directed towards 
the equator depends on the facts that the earth is not a true 
sphere, and that a continent consists of a floating mass of sial, 
the centre of gravity of which is higher than the centre of 
gravity of the displaced sima, or centre of buoyancy of the 
sial. Anywhere except on the equator and at the poles, a 
plumb-line on the surface of the earth points, not directly 
towards the centre of the earth, but to a part of the equatorial 
plane at a somewhat lesser depth than the centre. If the earth 
were a true homogeneous and non-rotating sphere, a plumb- 
line in latitude <, if produced downward, would make an 
angle < with the plane through the equator, but on the earth 
as actually constituted the angle would be greater than </, 
say, </>*. In a very deep mine the plumb-line, if produced, 
would pass nearer to the equator than if produced from the 
surface, that is, it would make with the plane through the 
equator an angle between <f> and <f>\ In a large mass of sial 
floating in a layer of sima, therefore, the downward force 
due to the attraction of the main mass of the earth on the sial, 
which can be considered as concentrated at the centre of 
gravity of the sial, is not exactly opposite in direction to the 


upward force due to the displaced sima, which can be con- 
sidered as concentrated at the centre of buoyancy. The 
resultant of these two forces is a small component towards 
the equator, which reaches its maximum in latitude 45 and 
vanishes at the poles and the equator. 

There is no doubt that such a force actually exists, and 
if the sima is a true fluid, however viscous, it would produce 
a slow movement of the continents towards the equator. 
The movement has been calculeited from the not very complete 
data available, and has been found to amount to 20 cm. 
(8 inches) a year in latitude 45, where it is greatest. But 
all this rests on the assumption that the sima really is a fluid, 
and that this fluidity persists even in the uppermost coldest 
layers. The equatorward force is so small that the resistance 
of a quite thin non-fluid layer at the surface would suffice to 
overcome it. Jeffreys (3, App. G.) is of opinion that while 
the earth may be considered to be a plastic body of zero 
strength at depths greater than 450 miles, there is some 
evidence that the cooler surface rocks have in fact a finite 
though small strength to depths of a few hundred miles. For 
instance, the floor of the ocean is strong enough to maintain 
the Tuscarora deep. This surface strength would probably 
be great enough to overcome the force due to the difference 
between the centres of gravity and of buoyancy. According 
to Jeffreys, then, it is doubtful whether the force available 
is strong enough to move the continents at all, and it becomes 
highly improbable that such a small force can raise enormous 
mountain-chains. To raise a slice of the earth's crust thousands 
of feet against gravity requires an enormous force, much 
greater than the small forces due to the difference between 
the values of gravity in different parts of the continents. The 
forces postulated by Wegener are of the order of one hundred- 
thousandth (io~ 5 ) of a dyne per square centimetre, whereas 
to elevate the Rocky Mountains a force of about one hundred 
million (io 9 ) dynes per square centimetre would be required. 
According to Jeffreys' calculations, therefore, the forces 
available on Wegener' s theory are about one hundred billion 
times too small for the effect which is attributed to them. 
Jeffreys sums up " Secular drift of continents relative to the 
rest of the crust ... is out of the question. A small drift 
of the crust as a whole over the interior, which would be 


shown as a displacement of the poles relative to the earth's 
surface, is not impossible, but the maximum amount seems 
too small to be of much interest." 

The forces which tend to produce motion in an east-west 
direction are much less clearly defined by Wegener than the 
force directed towards the equator ; they appear to depend 
mainly on the effects of tidal friction both on the floor of the 
sea and within the earth's crust, and to be of the same order 
of magnitude as the forces directed towards the equator. 
From the point of view of palaeoclimatology, the question of 
the east-west movement is of less importance than the question 
of the north-south movement. East-west movements of 
some of the continents relative to others may affect the annual 
range of temperature or the distribution of rainfall to some 
extent, but cannot radically change the mean temperature 
of the whole belt in any latitude, but rearrangements of the 
positions of the continents relatively to the poles can obviously 
be made to produce almost any changes of mean temperature 
which may be required to fit the evidence. 

This, in effect, is what Wegener does. Taking as a basis 
the power of free movement of the continents over the face 
of the earth, Wegener and Koppen (4) proceed to study the 
distribution of climates during the various geological periods, 
and to map out the positions of the continents relative to each 
other and to the poles which will best fit in with the distribution 
of climates, while preserving some continuity from one period 
to the next. It is assumed that the earth has been under 
solar control throughout, and that the distribution of climatic 
zones relative to the poles has always been similar to that 
found at present. On either side of the equator there has 
always been on the land a belt of rich vegetation represented 
by thick coal formations, and in the oceans a high-water 
temperature represented by reef-building organisms such as 
corals and Rudistes. On either side of this tropical belt there 
has always been over the land a zone of deserts. Nearer 
the poles another belt of vegetation in temperate latitudes has 
formed other coal beds less well developed, with annual 
growth rings in the tree stems. Finally, the sites of the poles 
have generally, if not always, been occupied by ice, either 
inland ice-sheets or floating ice-caps according as the pole 
lay on land or in the ocean. These zones at present do not 



run strictly parallel with the lines of latitude, and no doubt 
there were similar irregularities in the past, but Koppen and 
Wegener consider that these were never sufficient to mask the 
zonal distribution. It is proposed, first, to run briefly through 
the distribution of zones in the different periods according to 
this theory, reserving criticism to a later stage. 

The Late Carboniferous is the earliest period for which a 
good cartographical basis is available according to the con- 
tinental drift theory, but some attempt is made to reconstruct 
the earlier periods. Thus, in the Algonkian (Late Proter- 
ozoic), there was an intense glaciation of North America, 
which is therefore considered to have been the site of one of 
the poles, while the corresponding dry belt is represented by 
the Torridon Sandstone in Scotland and the Dala Sandstone 
in Central Norway. In the Early Cambrian [now believed 
to be mainly Late Proterozoic] there are more or less doubtful 
glacial deposits in Norway, Yangtse (China), South Australia, 
India (Salt Range), and South Africa. In Australia the 
glacial deposits are followed by thick limestones with reef- 
building Archaocyathina, indicating a rapid warming. The 
authors find that they cannot indicate the position of the poles 
and equator during the Cambrian and Ordovician periods. 

In the Silurian, the evidence is a little more definite, but 
orientation is still difficult. The occurrence of glacial deposits 
is doubtful, but there may have been ice in South Africa. 
The equatorial belt, represented by corals in the marine 
deposits and poor coal seams over the land, passed through 
North America, the British Isles, Central Europe, Northern 
Siberia, and possibly Australia. The northern desert zone 
lay near Leningrad, in Baffin Bay, and especially in North 
America. In the Devonian, there was still ice in South 
Africa, but Europe lay farther north than in the Silurian, 
the equatorial belt passing through France and Spain, Central 
Asia and China. Desert formations such as the Old Red 
Sandstone were extensively developed in the northern con- 
tinent, and the fauna includes the famous " lung-fish " 
(Ceratodus), and a " lung-snail. 55 This continent must therefore 
have " had a hot desert climate, whose dry periods were only 
occasionally interrupted by thundery rains. 55 
^ During the Carboniferous period the conditions are known 
in much greater detail, partly because of the interest aroused 


by the great glaciation which occurred during this period, 
and partly because the majority of the workable coal beds 
are of Carboniferous age. The greatest development of 
ice deposits occurred in South Africa, India, Australia, the 
Argentine and Eastern Brazil, and the Falkland Islands. 
These are so thick and extensive that they must be due to 
great inland ice-sheets, which Koppen and Wegener consider 
can only have formed in the neighbourhood of the poles. If 
we suppose the South Pole to have been in South Africa at 
that time, while the continents still had their present positions 
relative to each other, the most remote of them would still 
lie too near the equator to be readily glaciated. Hence the 
authors suppose that at this period Pangaea, the original 
continent, had not yet been split up into its component parts, 
and the glaciated continents were all grouped in contact with 
each other round the South Pole. The area covered by the 
ice-sheets was so extensive, however, that even this rearrange- 
ment does not suffice, and it is considered that the various 
glacial deposits are not all of the same age, but that glaciation 
followed the location of the moving pole. The Brazilian 
deposits are the oldest, the Australian and Indian the youngest. 
The South Pole travelled from Antarctica via South America 
to South Africa, and thence in a great arc across Australia 
back to Antarctica. The position of the equator is deter- 
mined by the gieat coal beds in North America, Europe, 
and China, and it is found that these lie on a great circle 
the centre of which falls in the glaciated region. These 
coal beds, therefore, represent the tropical rain-forest, a 
conclusion which will be discussed later. Other coal measures 
in Alaska, South America, South India, Australia, and 
Antarctica are attributed to the temperate rain belts. Between 
these temperate coal beds and the main mass of coal are a 
number of desert deposits. 

During the whole of the Mesozoic period there was little 
if any ice action, and the development of coal was also 
restricted. On the other hand, there was a great expansion 
of the dry belts, especially in North America and Africa. 
In the sea, great coral reefs were formed. The authors tacitly 
admit the generally accepted opinion that during the Mesozoic 
the development of climatic zones was less marked than at any 
other period since at least the middle of the Palaeozoic. It was 


during the Jurassic that the continents first began to drift 
apart, India and Antarctica splitting off from Africa, and 
Australia separating from Farther India. South America did 
not become an independent continent until the Cretaceous. 
Throughout the Mesozoic, the South Pole lay in Antarctica 
and the North Pole in the North Pacific. It may be remarked 
here that according to the reconstruction of Cretaceous 
geography the British Isles lay in about 20 N., and therefore 
had a tropical climate. The British chalk, however, contains 
a number of erratic pebbles, which are so alien to the general 
character of that deposit that it is difficult to attribute them 
to any other agency than floating ice, and this fact seems to 
be fatal to Wegener's reconstruction of that period. 

The interval of rest during the Mesozoic did not extend 
into the Tertiary, which was a period of great mountain- 
formation and of great and rapid shiftiiigs of the earth's axis, 
which brought the North Pole over the land and caused a 
great ice-age in the Northern Hemisphere. There are no 
certain traces of ice in the Eocene ; there was a considerable 
development of brown coal formation, especially in North 
America and Europe, but whereas the European beds are 
attributed to the equatorial rain belt, the North American 
beds are placed in the north temperate zone. (This should 
be compared with Berry's description of the Eocene floras 
quoted in the Introduction.) The Oligocene was generally 
similar to the Eocene, but in the Miocene we have the beginning 
of the ice-age in Alaska, North-east Siberia, and the New 
Siberian Islands. This is a very important point which is 
referred to again later. In the Pliocene, the Alaskan glaciation 
spread over the greater part of North America, including 
Greenland. At this time the North Atlantic existed only as a 
very narrow rift, and Greenland lay to the north or north-east 
of the British Isles. The east winds shown by the late F. W. 
Harmer to have blown across the North Sea are attributed to 
the glacial anticyclone associated with the American and 
Greenland ice-sheets. In the Miocene, the pole lay just 
north of Alaska. In the Pliocene it moved rapidly across the 
Canadian Arctic Archipelago, and at the beginning of the 
Pleistocene it lay near Disco Island off West Greenland. 
During the greater part of the Quaternary the North Pole 
lay in Central Greenland. In the Baltic Readvance it was 


near Spitsbergen, and then gradually assumed its present 
position. Ice-sheets were formed over the parts of the 
continents which lay nearest to the poles. The theory, 
therefore, requires a revision of the generally accepted 
correlation of the Quaternary deposits ; the Alaskan glaciation, 
as we have seen, is placed in the Miocene, along with a some- 
what hypothetical glaciation of British Columbia. In the 
United States the two earliest glaciations, Kansan and 
Jerseyan, are attributed to the Pliocene, the Illinioan is 
paralleled with the Gunz, the lowan with the Mindel, the 
Earlier Wisconsin with the Riss, and the Later Wisconsin 
with the Wurm. Correlation between American and Euro- 
pean glacial deposits is admittedly difficult, but this 
arrangement of the American sequence cuts right across the 
present opinions of most American geologists, which are set 
out on page 242. 

While the general course of the ice-age depended on the 
successive positions of the moving pole, the alternation of 
glacial and inter-glacial periods cannot be explained in this 
way, and the authors have recourse to the variations in the 
obliquity of the ecliptic and in the eccentricity of the earth's 
orbit. This part of the theory, and the objections to it, were 
dealt with in Chapter V. 

We may sum up the results of the investigation of climatic 
changes by Koppen and Wegener by giving the positions 
which they assign to the North Pole relative to the present 
position of Africa. Since Europe has always had almost its 
present position in relation to Africa, these figures give also 
the various positions of the North Pole relative to Europe. 

Latitude of 
Period. Position of North Pole. England. Antarctic. 

Carboniferous . . 30 N. 145 W. o 75 S. 

Permian .... 35 N. ii5W. 15 N. 70 S. 

Trias 50 N. 125 W. 20 N. 85 S. 

Jurassic .... 47 N. 132 W. 20 N. 90 S. 

Cretaceous . . . 47 N. 140 W. 15 N. 85 S. 

Eocene 45 N. 160 W. 15 N. 90 S. 

Miocene .... 75 N. 150 W. 40 N. ca. 90 S. 
Late Pliocene and 

Early Pleistocene 70 N. 60 W. 60 N. ca. 85 S. 

Table 18. Changes of latitude according to Wegener. 


In the third column I have added the corresponding latitude 
of England, and in the fourth column the latitude of the 
centre of the Antarctic continent. 

The zonal arrangement of climates has persisted throughout 
geological time, and though there have probably been minor 
fluctuations in the average rainfall over the globe, the 
variations of climate in any one area have been governed almost 
entirely by the variations of latitude which it has undergone. 
In the next chapter we will examine the latter contention in 
greater detail. 


(1) WEGENER, A. " The origin of continents and oceans." Transl. by J. G. A. 

Skerl. London, 1924. 

(2) CLOSE, SIR CHARLES. " The geodetic evidence for the supposed westerly 

drift of Greenland." London, Geogr. J., 63, 1924, p. 147. 

(3) JEFFREYS, H. " The earth, its origin, history, and physical constitution." 

2nd ed. Cambridge, 1929. 

(4) KOPPEN, W., UND A. WEGENER. " Die Klimate der geologischen Vorzeit." 

Berlin, 1924. 



WE will begin the critical discussion of the views set 
out in Koppen and Wegener's book, " Die Klimate 
der geologischen Vorzeit," with some further analysis 
of the climatic conditions during the Upper Carboniferous 
period, beginning with the United States, the British Isles, 
and Central Europe. According to the " drift " theory, 
these coal beds represent a luxurious tropical rain-forest, 
and the equator is therefore drawn as nearly as possible through 
the middle of them. The evidences of glacial action which 
have been adduced from time to time in close proximity, both 
in space and time, to these coal beds are dismissed out of hand 
as not genuine. The American evidence, however, seems to 
be too well founded to be dealt with in this summary fashion. 
Thus S. Weidmann (i) describes conglomerates of Upper 
Carboniferous to Permian age in the Arbuckle and Wichita 
Mountains of Oklahoma and in Kansas, associated with all 
the paraphernalia of glaciation scratched boulders, erratics, 
fluted and polished floors, and U-shaped valleys. Some of 
the boulders in marine deposits have apparently been carried 
by icebergs, and the author attributes the phenomena to 
islands in the Late Palaeozoic sea bearing local valley glaciers. 
J. A. Taff (2) found boulders up to 20 feet across and 5 or 6 
feet thick, 50 miles or more from their source, in the marine 
Caney shales of Eastern Oklahoma. " No other competent 
means of their transportation than ice presumably heavy 
shore ice has been suggested. 55 Similarly, A. P. Coleman 
(3) considers that there is good evidence for glaciation in 
Oklahoma, Nova Scotia, and Alaska (Thousand Isles). As 
regards Nova Scotia, Coleman writes, u It is ... probable 
that there were moderate elevations from which, under a cool 
climate, glaciers spread out over the plains on which coal 
forests had been growing not long before. 55 


In the Squantum tillite near Boston, Mass., there are 
massive conglomerates 2,000 feet in thickness, which cover a 
considerable area. The chief interest of these beds, apart 
from the presence of striated boulders, lies in the associated 
" varve " beds (4) banded clays which are similar in all 
respects to those formed during the retreat of the Scandinavian 
and North American ice-sheets at the close of the Quaternary 
glaciation, and also similar to those formed in Australia during 
the Upper Carboniferous glaciation. These clays owe their 
banding to the seasonal variations in the rate of melting of 
glaciers, and are therefore incompatible with an equatorial 
climate. In places the banding is disturbed during the 
deposition of the shales, probably by the grounding of floating 

Wegener recognises that the Squantum tillites demand 
serious consideration, and the effort he makes to explain 
them away tacitly implies that they form a very serious 
objection to the " continental drift " theory. He considers 
the possibility that they are real, but were formed at a very 
high level in the Appalachian mountain system, then young 
and vigorous, and agrees with the general opinion of geologists 
that the chances of preservation of high-level glacial deposits 
over a wide area would be very slight. As we have seen, there 
are other indications that the glaciers reached low levels. 
He therefore concludes that as smoothed floors have not been 
found beneath the moraines, the remaining phenomena, 
although very suggestive of glacial action, could have 
originated in other ways. As all the other evidence indicates 
that Boston lay in the equatorial rain zone during the 
Carboniferous and in the region of hot deserts during the 
Permian, " the glacial nature of these tillites is in irreconcilable 
opposition to the numerous climatic traces of another kind which 
surround it both in space and time." He therefore says that the 
burden of proof that the deposits are really glacial rests with 
the opponents of the " drift " theory. The Squantum tillites 
are, however, accepted by all American geologists, while 
the Caney shales in the Arbuckle and Wichita Mountains 
were examined by an impartial observer, J. B. Woodworth, 
who concluded (5) that while the striae on the boulders which 
he observed were probably not glacial, the transport of the 
boulders was almost certainly effected by floating ice, 


Farther west, in Colorado (6), there are Middle Carbon- 
iferous conglomerates some 6,000 feet in thickness, said to 
contain boulders up to 50 feet in diameter, and although no 
striated blocks have yet been found, the great size of the 
boulders strongly suggests ice action. 

C. A. Slissmilch and Sir T. W. E. David (6) discuss in 
detail the deposits in Europe which suggest ice action. The 
evidence, while not so strong as that from North America, 
yet has some interest. Sir Andrew Ramsay was of opinion 
that glaciated pebbles occurred in the Permian conglomerates 
of England, but this interpretation is not now accepted. The 
Millstone grit, although it contains no striated material, 
points to enormous denudation. In France, M. Julien has 
described large masses of angular breccia in the St Etienne 
coal basin, with a thickness up to 800 feet. Striae are extremely 
rare, but some have been found. " Vertical roots of 
Calamites are seen in the sandstones underlying the breccias, 
while their stems, as they pass upwards into the breccia, are 
crushed, a phenomenon very suggestive of glacial action. . . 
He considers that these c morainic breccias ' were deposited 
by glaciers having their origin in the great early formed folds 
of the Hercynian ranges which were already rising to the 
north." In Germany there is also some rather doubtful 
evidence of glacial action, the strongest being a shale bed con- 
taining occasional boulders up to a foot or more in diameter, 
suggesting the action of shore ice or river ice. The European 
evidence, taken by itself, is not very convincing, but in con- 
junction with the much stronger American evidence it throws 
considerable doubt on the theory that the coal measures aie 
the remains of equatorial rain-forests. 

So great an authority as A. P. Coleman (7) examined the 
distribution and sequence of both Permo-Carboniferous and 
Pleistocene glacial deposits from the point of view of the 
continental drift hypothesis, and concluded that the latter 
fails completely to account for them. The extent of the 
Permo-Carboniferous glaciated area was so great that even 
if the continents were joined up round the South Pole ice- 
sheets would still extend into sub-tropical latitudes. There 
is, however, evidence that in all the glaciated areas the ice 
reached the sea, and in South America and Australia there 
was open sea on both sides of the continent. In any case 


such a gigantic continent as that postulated by Wegener 
would be highly unfavourable to glaciation owing to the 
difficulty in the supply of moisture. 

There is one other piece of evidence in connexion with 
the climatic zones of the Permo- Carboniferous which may 
be referred to, not so much for its intrinsic importance, 
which is small, as because it illustrates the methods too often 
adopted by Koppen and Wegener in dealing with items which 
do not quite fit their theory. Salt beds occur in Angola, 
formerly attributed to the Carboniferous. The authors class 
these as Permo-Triassic, because during the Carboniferous 
" Angola was too near the South Pole for salt beds to form." 
The drift hypothesis has certainly not reached a stage of proof 
in which it can be asserted that evidence which does not fit 
it is thereby proved to be false. 

The next point in the discussion of the " drift " theory 
concerns the relative ages of the ice-sheets in the Southern 
Hemisphere. According to Koppen and Wegener, the 
Brazilian deposits are the oldest, the Australian and Indian 
the youngest, and L. Waagen is quoted as the authority for a 
statement that the glaciation of Brazil and South Africa 
occurred before the development of the Glossopteris flora, that 
of Australia after it. The age of the glacial deposits in different 
parts of the world is discussed in great detail by Sussmilch 
and David (6), and they arrive at very different conclusions. 
The succession in New South Wales is taken as the standard 
of reference, and from the generalised section given, we may 
make out the following simplified series : 

p ^ . M ( Glossopteris coal measures, etc. 

JL ermian ..... x /-+ i i i i 

Crmoidal shales. 

Branxton glacial horizon. 

Greta coal measures. 


Basalt and tuff. 

Horizon of Eurydesma cordatum. 

Upper Carboniferous 

Gangamopteris muds tone. 

Brandon conglomerate. 

Tillites, etc. 

Varve beds. 

Rhacopteris horizon. 

Tuffs, fluvio-glacial conglomerates, etc. 


There are thus two main glacial series, the older and more 
important falling in the lower part of the Upper Carboniferous, 
while the upper horizon falls probably at the top of the Upper 
Carboniferous. Between the two glacial series we find 
Gangamopteris and Eurydesma, while Glossopteris first occurs 
above the upper glacial horizon. In Victoria, Tasmania, and 
South Australia the main glacial horizon lies below the 
Gangamopteris horizon, and probably on the same horizon as 
the lower tillites of the New South Wales series. Tillites on 
the Irwin River in Western Australia are probably somewhat 
younger, falling in the Upper Carboniferous near the zone of 
Eurydesma cordatum. In Western Australia there are two 
glacial horizons, both older than the Glossopteris flora. 

In India (Salt Range) the tillites are associated with 
Enrydesma, and may be referred to the Upper Carboniferous. 
In South Africa the Dwyka tillites are likewise associated 
with Eurydesma, and the thick series probably belongs mainly 
to the Upper Carboniferous. In the Falkland Islands the 
boulder beds are Upper Carboniferous and occur beneath 
beds containing Gangamopteris and Glossopteris., and appear to 
come at the base of the Per mo- Carboniferous. Finally, the 
South American tillites occur just beneath coal measures with 
an exclusively Gangamopteris flora, and may be attributed to 
Upper Carboniferous. There was very little glaciation in the 
Lower Permian anywhere. 

Thus the evidence, as set out in 1919 by two unbiassed 
observers, and still further emphasised by Sir T. W. Edgeworth 
David in a series of lectures in 1926, is to the effect that there 
was very little difference in age between the main glaciations 
of the different areas in the southern group. All of them are 
certainly older than the Glossopteris flora, and the only feature 
which supports the theory of a moving pole is the slight 
recrudescence of glaciation in the highest Carboniferous or 
possibly lowest Permian of New South Wales (Bolwarra and 
Branxton beds). 

Sussmilch and David note the close relationship between 
the areas of folding and the glaciated areas, and also the 
enormous thicknesses of volcanic tuff. They further note 
that the period of glaciation of Eastern Australia was almost 
exactly synchronous with the period of great orogenic move- 
ments, also in Eastern Australia, and they suggest that the 


glaciations were due largely, if not entirely, to the presence of 
high mountains and of great quantities of volcanic dust. 

The recognition of glacial deposits and of ice floating in the 
sea in the middle of the northern belt of coal measures, pro- 
foundly modifies the problem of the habitat of coal-building 
plants. For a long time it was believed that peat, which 
is the first stage in the formation of coal, could not form in a 
hot country owing to the rapidity of decomposition at high 
temperatures, and that the modern representatives of the coal 
measures are the peat-bogs of moist temperate regions. The 
reversal of this view is mainly due to the influence of H. 
Potonie, who described a swamp in Eastern Sumatra in which 
peat is actually being formed at the present moment, almost 
entirely from the fallen leaves of evergreen trees. Koppen 
and Wegener consider that the coal beds extending through 
the Eastern United States, the British Isles, Central Europe, 
and China represent the remains of similar peat formed by 
tree ferns and other highly developed vegetation in equatorial 
swamps of the Carboniferous period, while the coal beds of 
Spitsbergen in the north, and of Australia, South Africa, 
South America and Antarctica in the south, represent the peat 
of the temperate rain belts. It seems very doubtful whether 
we are entitled to draw this conclusion solely from the nature 
of the vegetation composing the coal beds. The absence of 
annual rings of growth in the Carboniferous vegetation of the 
north temperate belt may be due, as E. Antevs (8) points out, 
to the comparatively low organisation of the flora, which only 
formed annual rings under extreme conditions. If, as 
suggested in Chapter V., the coal beds of the Northern 
Hemisphere were formed during a period of high eccentricity 
of the earth's orbit, at the times when northern winter was 
in perihelion, we need not expect to meet annual rings until 
we reach very high latitudes, such as Spitsbergen. Through- 
out the Lower and Middle Carboniferous, almost to the 
beginning of the ice-age in Australia, the flora was extra- 
ordinarily uniform over the whole world (9), Europe, Asia, 
Africa, America, Australia all had the same flora. The 
Glossopteris flora is of later date than the northern coal beds ; 
it represents a different stage in the evolution of plant life, 
but not necessarily a difference of climate. This flora was 
able to spread across the equator into the Northern 


Hemisphere, a fact which suggests a higher organisation than 
that of the older flora rather than a special adaptation to cold 
climates. The position is similar to that at the end of the 
Mesozoic, when a flora of modern type originated in high 
northern latitudes and spread over the world, and in the latter 
example there is no suggestion of a sweeping change of latitude. 
The distribution of desert deposits salt, gypsum, and 
desert sandstone also requires a closer examination. In 
Fig. 26 the occurrence of these indications of desert action in 

D desert sandstones. S salt. G gypsum. Dotted area present deserts. 

Fig. 26. Carboniferous arid Permian deserts in relation to the 
present desert areas. 

the Carboniferous and Permian periods is shown by their 
initial letters S, G, and D. The distribution of present deserts 
is indicated by stippling. It is seen that with the exception 
of a few salt deposits in Europe, Western Asia, and the east 
of North America, the late Palaeozoic desert deposits in the 
Northern Hemisphere fall entirely in the present desert areas. 
The present deserts of the Southern Hemisphere were, however, 
unrepresented in the Carboniferous and Permian. This 
presumably indicates that during those periods the present 
Southern Hemisphere was moister than at present, either 
because of the shifting of the continents relative to the poles, 
or because the existence of the ice-sheets caused a " pluvial 
period >J in the surrounding lands. 

These considerations show that the theory of " continental 
drift " is not so complete and irresistible an explanation of the 
peculiar distribution of climate in the Carboniferous and 


Permian periods as Koppen and Wegener seem to think. 
The greatest difficulty is presented by the glacial deposits of 
North America, especially the " varve clays," and these form 
almost as great an obstacle to the " drift " theory as the glacial 
deposits of equatorial Africa form to the assumption that the 
continents held their present positions. The only complete 
answer to the " drift " theory, however, would be a demons- 
tration that the climatic events of the Carboniferous and 
Permian were the logical result of the distribution of land, 
especially high land, and sea during that period, the poles 
being supposed to have kept their present positions. 

In the Mesozoic and Tertiary periods the discrepancies 
between the past and present distribution of climatic zones 

D desert sandstones. S salt. G gypsum. Dotted area present deserts. 

Fig, 27. Mesozoic deserts in relation to the present desert areas. 

are far less striking. There was little ice, except towards the 
close of the Tertiary, and on any theory the climatic zones 
were much less developed than at present. It is not possible 
to discriminate between equatorial and temperate coals, and 
hence the desert deposits offer practically the only evidence 
of zoning. But as Fig. 27 shows, the distribution of deserts 
during the Mesozoic was very similar to the present distribution. 
The chief exceptions were in Europe and the South-eastern 
United States. The rainfall of Europe would be greatly 
decreased by the displacement of the Icelandic minimum 
far to the north, while the present comparatively heavy rainfall 
of the South-eastern United States is due almost entirely to 


moisture derived from the warm Gulf of Mexico. In fact, 
the distribution of deserts shown in Fig. 27 is very nearly 
what we should expect during a period in which low continents 
and a free oceanic circulation reduced the temperature 
gradient between low and high latitudes almost to its least 
possible value. The distribution of Rudistes in the Cretaceous 
period is the strongest point in favour of the displacement of 
the poles. Reef-building forms reach their greatest develop- 
ment in the deposits of the great Tethys Sea, the Mediterranean 
of the Cretaceous, which, in contrast with the present Mediter- 
ranean, was open to the Indian Ocean and received a constant 
supply of warm water. They are also found in the north of 
South America, Mexico, and the West Indies, the range 
extending to the north coast of the Gulf of Mexico, the limits 
in America so far discovered being 5 and 30 N. The most 
interesting point is that outside the normal range dwarf forms 
are found, in latitudes 50 to 55 N. in Europe, but in latitudes 
5 to 20 S. in East Africa. If the distribution of Rudistes was 
really governed by the sea temperature and we have no 
reason to suppose that it was not the assumption of the 
present position of the continents requires a much greater 
northward displacement of the thermal equator than occurs at 
present, and especially a cold current along the east coast 
of Africa instead of along the west coast (see Fig. 28) . Probably 
the only cause which could produce such a great displacement 
of the thermal equator would be an extensive glaciation of the 
Antarctic continent, while the north polar regions were prac- 
tically free of ice. The glaciation of the Antarctic continent, 
according to C. S. Wright and R. E. Priestley (10), began at 
least as early as the beginning of the Tertiary, since immediately 
above the Cretaceous beds at Cape Hamilton, Graham Land, 
there occurs a "moraine-like mass, some metres in thickness," 
which contained angular fragments of crystalline rocks foreign 
to the locality. This deposit is evidently glacial, though it 
may mean nothing more than a local valley glacier ; it 
cannot be younger than Eocene and may be uppermost 
Cretaceous. The onset of cold conditions during the 
Cretaceous is also indicated by the Cockburn Island sandstone, 
which was " crowded with pygmaean forms of life such as 
might be expected to result from the encroachment of colder 
conditions upon a marine fauna long developed in, and 




habituated to, more genial conditions." There was shore ice 
or perhaps icebergs in the Cretaceous sea over South Australia, 
the erratics covering a wide area. 

In the Northern Hemisphere there is no direct evidence of 
the occurrence of extensive ice-sheets during the Cretaceous. 
There is, however, some evidence of drift ice in the remarkable 
erratic blocks found in the English upper chalk. These blocks 
are probably due to transport by shore ice rather than by 
icebergs, but in any event their presence is a serious obstacle 
to the belief that at that time England lay in about 20 N. 
latitude. The problem is very similar to that of the Upper 
Carboniferous, since in both periods the biological evidence 
points to high temperatures, while the character of the deposits 
indicates ice action in the neighbourhood. 

The most complete refutation of the drift hypothesis is 
given by the fossil plants of the Mesozoic and early Tertiary. 
E. W. Berry (n) states that " the distribution of the known 
fossil Arctic floras with respect to the present pole proves 
conclusively that there could have been no wandering pole." 
R. W. Chaney (12) gives maps of Eocene " isoflors " in the 
northern hemisphere, showing the limits of sub-tropical, 
temperate and cool temperate floras as irregular lines 
surrounding the pole. These enclose elliptical areas which 
approach the pole most closely round about longitude o and 
1 60 W., z.., in the openings between the circum-polar 
continents. The limit of the " cold-temperate " flora has an 
average latitude of about 78 N. centred four degrees from the 
pole. The limit of the " temperate " flora has an average 
latitude of 62 N. and is also centred 4 from the pole, that 
of the sub-tropical flora has an average limit of 42 N. 
and is centred 3 from the pole. The displacements of 
the first two are towards Bering Strait, that of the third 
towards Asia. The present mean annual isotherm of 40 F. 
is also displaced about 4 from the pole in the direction 
of western Siberia. The agreement could hardly be better, 
and practically amounts to proof that in the Eocene the 
North Pole occupied its present position and not a point in 
the North Pacific as shown by Wegener. 

The correlation of the glacial stages during the last ice-age 
in America and Europe according to Koppen and Wegener 
differs from that generally accepted. The most important 



point deals with the date of the main glaciation in Alaska, 
which Koppen places in the Miocene. W. H. Dall, the 
representative of the American Geological Survey in Alaska, 
does not accept this correlation as possible. He gives the 
following sequence at Nome (13) : coal formation during 
Eocene and Oligocene, locally covered by marine fossiliferous 
Miocene, indicate a mild climate in the early Tertiary. During 
the Miocene, the land sank for the most part below sea-level, 
and there was much volcanic activity ; the climate became 
cool temperate in the Early and Middle Miocene, but warmed 
up again in Late Miocene. Since the Miocene, the land has 
risen continuously. In the Pliocene the climate was moderate, 
and it was not until the Quaternary that Arctic temperatures 
set in, to persist until the present day. 

The direct correlation of glacial deposits in North America 
and Europe is difficult, and we have to rely mainly on such 
features as the determination of ages by the relative depths 
to which the various deposits have been weathered. A careful 
study of all lines of evidence has recently been made by 
Osborn and Reeds (14), who accept in the main the results of 
F. Leverett (15), based on a comparison of the depth of 
weathering of glacial deposits and on the texture and fauna 
of the loess. This correlation gives : 

I. Glaciation Gunz, Scanian, Nebraskan, Jerseyan. 

1 . Interglacial Gunz - Mindel, Norfolkian, 


II. Glaciation Mindel, Kansan. 

2. Interglacial Mindel-Riss, Yarmouth. 

III. Glaciation Riss, Illinoian. 

3. Interglacial Riss-Wurm, Sangamon. 

IV. Glaciation Wurm, Wisconsin. 

This correlation is in fact almost inevitable. In Europe, 
the Mindel-Riss interglacial is distinguished from the Gunz- 
Mindel and the Riss-Wurm by its much greater length, and 
by a temperature which probably rose higher than the present 
temperature, by being in fact an " interglacial " rather than 
an " intraglacial " period. Similarly, in America, the interval 
between the Kansan and Illinoian was much greater in length 
than the remaining interglacial periods. Since the work of 


de Geer and Antevs has shown that the latest glaciations in 
North America and Sweden were contemporaneous, this 
estimation of the age of glacial deposits by the depth of 
weathering seems to require that the long Yarmouth stage 
be correlated with the long Mindel-Riss interglacial. The 
astronomical correlation of the glacial stages has been dis- 
cussed in Chapter V. ; it is not an essential part of the theory 
of continental drift and need not be referred to again here. 

It is always a useful test of a new theory to examine how 
far facts which come to light after the theory has been completed 
fit into place. An opportunity for such a test is afforded by 
the study of the climatic history of Antarctica given by Wright 
and Priestley in the volume of results of the British Antarctic 
expedition dealing with glaciology (10). According to 
Koppen and Wegener's reconstructions, Antarctica as a whole 
has been in high latitudes since at least the beginning of the 
Carboniferous period. Of course the Antarctic continent is 
large, and if the pole was at one side of the continent the 
opposite side would extend into temperate latitudes, so I 
have tried to pick out on Wegener's charts the particular 
point to which Wright and Priestley's climatic indications 
refer. In this way I have obtained Table 1 9 (see page 244) . 

From this table we see that during the Upper Carboniferous 
Antarctica had more or less the climate appropriate to its 
latitude according to Wegener. In the Permian and Triassic, 
however, while the continent was drifting into continually 
higher latitudes, the climate was steadily ameliorating. The 
flora of the Jurassic rocks of Graham Land is extremely rich, 
and closely resembles that of the Jurassic rocks of Europe, 
which according to Wegener then lay in about 15 N. latitude. 
The Cretaceous fauna, while still rich, contains at least one 
bed of dwarf forms suggesting the oncoming of colder con- 
ditions. The only determinable plant fossil, a Sequoia, finds 
its nearest relative in the Cretaceous of Europe and Greenland. 
It is evident that the climate of Antarctica throughout the 
Mesozoic was quite incompatible with its latitude according 
to Wegener's theory, unless we assume in addition a great 
amelioration of the polar climate. For the Mesozoic, there- 
fore, the theory of continental drift presents no advantage over 
any other theory. The same applies to the warm climate 
of the Oligocene. The doubtful glaciation of the Eocene 



Carbon ifcrous . 

Permian . 








South Victoria 


Adelie Land. 
Graham Land. 

Graham Land. 


Temperate to hot or 
cold desert. No 
definite evidence of 
glacial conditions, 1 
but strong evidence 
of seasonal climate. 

Sub-tropical to warm 

Temperate to warm 

Graham Land. First Glariation (?). 

Seymour Island. 
Cape Hamilton. 
Cape Adare. 

Campbell Island. 
Cockburn Island. 

Sub-tropical to temper- f 
ate becoming frigid. J 

Temperate (?). 

Maximum extension of 



to Wegener. 

65 S. 

75 S- 

85 S. 

70 S. 

75 S. 

80 S. 

50 s. 

65 S. 

Table 19. Variations of climate in Antarctica. 

is not very good evidence ; from the description given it 
resembles the moraine of a mountain glacier descending a 
narrow valley, and it is certainly of far less importance than 
the Carboniferous glacial deposits of North America. Campbell 
Island is not really Antarctic at all, and the cold climate of the 
Pliocene and Quaternary would be expected on any theory. 

We may also compare the climatic history of Australia 
according to C. A. Sussmilch (16) with its position according 
to Koppen and Wegener. Taking the mean of their figures 
for Perth, Cape York and Hobart, we find that the present 
latitude is 29 S. In the Carboniferous the latitude of 
Australia was 59 S. and the climate, warm at first, became 
very cold by Mid-Carboniferous. In the Permian the 
latitude was 71 and the climate cold at first, becoming 

1 Wegener fits the Antarctic continent into the Great Australian Bight. 
According to Sir T. W. Edgeworth David, the Australian ice-sheet radiated from 
a point to the south-west of Tasmania, which according to Wegener 's recon- 
struction would be in the Antarctic. The non-glaciation of the Antarctic in the 
Upper Carboniferous, if confirmed by further research, will be a strong point 
against Wegener 's reconstruction. 


warmer. So far the agreement is satisfactory. In the 
Trias the latitude was 68 and in the Jurassic 65, but 
according to Siissmilch the climate was probably warmer than 
to-day. In the Cretaceous Australia had moved north, to 
57, but the climate was colder. In the Miocene the latitude 
was 43 but the climate was at least 10 F. warmer than 
to-day. In the early Quaternary Australia moved to 54 S. 
and the climate became colder, but not very cold. There 
is in fact no agreement between the climatic changes in 
Australia and the path of the South Pole according to Koppen 
and Wegener. 

The evidence of the pre- Carboniferous deposits is still 
very meagre. The constitution of the slate-greywacke 
formation of Robertson Bay, South Victoria Land, which is 
of Late Proterozoic or very early Palaeozoic age, strongly 
suggests the action of alternate freezing and thawing, and 
these deposits may be the Antarctic representatives of the 
Late Proterozoic-Early Cambrian glaciation. Later in the 
Cambrian we have evidence of a moderately warm sea 
stretching nearly or right across Antarctica, in the form of 
thick limestones very rich in reef-building Archtfocyathina. 
Compared with the forms from Australia, however, all the 
Antarctic forms are either embryonic or dwarfed, indicating 
that they lived in colder and presumably more southern seas. 
No reliable evidence is yet available from the Silurian or 
Devonian periods. It seems, therefore, that the chief, perhaps 
the only, justification for the theory of continental drift rests 
on the distribution of climatic zones during the Upper 
Carboniferous period. 


(1) WEIDMANN, S. " Was there a Pennsylvanian-Permian glaciation in the 

Arbuckle and Wichita mountains of Oklahoma?" J. Geol., Chicago, 
31, 1923, p. 466. 

(2) TAFF, J. A. " Ice-borne boulder deposits in Mid-Carboniferous shales." 

Bull. Geol. Soc. Amer., 20, 1908, p. 701. 

(3) COLEMAN, A. P. " Late Palaeozoic climates." Amer. J. Set., 9, 1925, p. 195. 

(4) SAYLES, R. W. " Seasonal deposition in aqueoglacial sediments." Cam- 

bridge, Mass., Mem. Mm. Comp. <W., 47, 1919, No. i. 

(5) WOODWORTH, J. B. " Boulder beds of the Caney shales at Talihina, 

Oklahoma." Bull. Geol. Soc. Amer., 23, 1912, p. 457. 

(6) SUSSMILCH, C. A., and T. W. E. DAVID. " Sequence, glaciation, and 

correlation of the Carboniferous rocks of the Hunter River district, New 
South Wales." Sydney, J. R. Soc. N.S. Wales, 53, 1919, p. 246. 


(7) GOLEMAN, A. P. " Ice ages and the drift of continents." J. Geol., Chicago, 

4*> 1933, P- 4<>9- 

(8) ANTEVS, E. " The climatologic significance of annual rings in fossil woods. 

Amer. J, Sri., 9, 1925, p. 296. 

(9) ZEILLER, R. " Les provinces botaniques de la fin des temps primaires." 

Rev. gen. sciences, 8, 1897, p. 5. 
(10) BRITISH (TERRA NOVA) ANTARCTIC EXPEDITION, 1910-1913. " Glaciology," 

by C. S. WRIGHT and R. E. PRIESTLEY. London, 1922. 
(i i) BERRY, E. W. " The past climate of the North Polar region." Washington, 

Smithson. Misc. Coll., 82, 110. 6, 1930. 

(12) CHANEY, R. W. "Tertiary forests and continental history." New York, 

Bull. Geol. Soc. Amer., 51, 1940. 

(13) DALL, W. H. " Pliocene and Pleistocene fossils from the Arctic coast 

of Alaska and the auriferous beaches of Nome, Norton Sound, Alaska." 
U.S. Geol. Survey, Prof. Papers, 125 C., 1920, p. 23. 

(14) OSBORN, H. F., and G. A. REEDS. " Old and new standards of Pleistocene 

division in relation to the pre-history of man in Europe." Bull. Geol. 
Soc. Amer., 33, 1922, p. 411. 

(15) LEVERETT, F. " Comparison of North American and European glacial 

deposits." s. Gletscherk., 4, 1910, p. 241. 

(16) SUSSMILCH, C. A. " The climate of Australia in past ages." J. Proc. R. 

Soc. N.S. Wales, 75, 1941, p. 47. 



WE have now to look more closely at the geographical 
and climatic conditions of the Upper Carboniferous, 
in order to see if the distribution and eJevation of 
the land-masses were such that ice-sheets might conceivably 
have developed in low latitudes, while at the same time a 
comparatively mild climate obtained farther north. Fig. 29 
gives a rough reconstruction of the geographical conditions on 
the supposition that the continents were in their present 
positions. This reconstruction is based on that given by 
Th. Arldt (i) with some alterations to include the results of 
later work. In the Northern Hemisphere we find three small 
continents : Nearctis, a primitive North American continent ; 
North Atlantis, including Greenland and Western Europe ; 
and Angaraland, occupying part of the present Siberia. 
Nearctis and North Atlantis were connected by a land-bridge 
in about latitude 50 N. South of these three continents the 
Tethys Sea, the forerunner of the Mediterranean, extended 
east and west from New Guinea to Central America, sending 
an arm between North Atlantis and Angaraland to the Arctic 
Ocean. This Tethys Sea was bounded on the south by the 
great continent of Gondwanaland, extending in a huge 
irregular crescent from South America to Australia, and from 
20 N. to 40 S. In connexion with the great extent of 
Gondwanaland from west to east, it may be remarked that the 
Upper Carboniferous marine fauna of South-eastern Australia 
resembles that of South Africa, while the fauna of Western 
Australia is quite different and resembles that of the Tethys 
Sea. This indicates that there was a continuous land barrier 
separating the gulf west of Australia from the seas south-east 
and east of Africa. 

The principal difference between the land and sea dis- 
tribution of the Middle and that of the Upper Carboniferous 
seems to be that in the former Gondwanaland was not 





continuous from South America to Australia, but was probably 
broken up into three or perhaps four separate land-masses 
by straits leading from north to south. These are indicated 
by the broken lines of Fig. 29. By allowing free circulation 
between the waters of the Tethys Sea and those of the Southern 
Ocean, these breaks in the land barrier would raise the tem- 
perature of the Southern Hemisphere considerably, and help 
to account for the great climatic difference between the 
Middle and Upper Carboniferous. 

We know that the Carboniferous was a period of great 
mountain-building. The mountain ranges followed two 
main directions, north-south and east-west. A range followed 
the south coast of North Atlantis into the Mediterranean 
region. The site of the Alps was occupied by the Carnic 
range of Mont Blanc and there were other ranges in the 
Caucasus and the Dobrudja. In the west, the Pyrenees and 
Asturia were mountainous. Farther north a range or a series 
of ranges ran from Bohemia first northward through the 
Sudetes and then westward through Germany. In the west 
we have the Armorican chain through Brittany, South 
England, Wales, South Ireland, and beyond into North 
Atlantis. In North America the Appalachian Mountains 
were forming, turning westwards in the south to the mountains 
of Oklahoma and South Arkansas. Other north-south 
ranges formed the west coast of Nearctis, from Colorado 
northwards. In Asia there were a number of east-west ranges 
somewhat similar to the present systems, bounded on the 
west by the Urals and the Volga Sea, on the south by the 
Tethys Sea. The mighty continent of Gondwanaland was 
bounded on the west by the Proto-Cordilleras and the 
mountains of the Pampas, extending to the Falkland Islands. 
In the south of Africa there was another chain, and a great 
range ran along the whole eastern coast of Australia and the 
present East Indies. These mountain systems are shown by 
the heavy lines in Fig. 29. Between the mountain ranges in 
the Northern Hemisphere there were, for the most part, wide 
moist valleys open to the sea, the home of a rich vegetation. 

Opinions differ about the structure of the main area of 
Gondwanaland, i.e., whether it consisted of an extensive 
high plateau or a series of mountain ridges. It is generally 
agreed, however, that the Upper Carboniferous was a period 


of great mountain-building and the general elevation was 
probably high. The great thickness of the Upper Carbon- 
iferous of South Africa, for example, points to rapid 
denudation, suggesting a large area of high ground in the 
interior of that continent. The fact that the ice-sheets 
spread out from a line near the equator shows that initially 
at least the ground was highest there, and may well have been 
a ridge 10,000 or 15,000 feet above sea-level. 

Finally, it has to be remarked that the Upper Carboniferous 
was a time of intense volcanic activity, and especially in 
Australia, great thicknesses of agglomerates point to numerous 
explosive eruptions from which we may infer the presence of 
great quantities of volcanic dust in the atmosphere, forming 
a veil which, as Humphreys has shown, would be very effective 
in shutting out the solar radiation, while it would allow the 
terrestrial radiation to escape with little hindrance (see 
Chapter VI.). 

At the beginning of the Upper Carboniferous there appears 
to have been a general decrease in the temperature of the 
seas, indicated by an impoverishment of the fauna and flora 
resulting from the extinction of a number of animals and the 
withdrawal of the coral boundaries towards the equator. 
The land plants also suffered changes, and the introduction 
of holometabolism in insects (i.e., the pupa stage) is attributed 
by A. Handlirsch (2) to this decrease of temperature. Sir 
T. W. Edgeworth David considers that the mean temperature 
of the tropical oceans decreased by about 10 F., and it 
is probable that the oceans were ice-covered in high latitudes. 

We can now attempt to reconstruct the distribution of the 
meteorological elements. Over the open Pacific Ocean there 
is no reason to suppose that the system of pressure and winds 
was appreciably different from that prevailing now. Hence 
we postulate a great equatorial current setting westward 
towards the eastern coast of Gondwanaland, with a temperature 
of about 70 F. The configuration of this coast of Gond- 
wanaland appears to have been very favourable for concen- 
trating the warm current and directing it into the Tethys Sea, 
more favourable even than the present configuration of the 
coast of America for concentrating the equatorial currents of 
the Atlantic in the Northern Hemisphere. A very warm 
current, with a temperature initially in the neighbourhood 


of 70 F., must have flowed through this narrow inter- 
continental sea. The supply of warm water would have been 
large enough to give this warm current a considerable velocity 
perhaps fifty miles a day enabling it to conserve its heat 
over a long distance. Part of this warm current turned 
northward through the Volga Sea and brought favourable 
conditions to Northern Russia, where the Fusulinse appear 
to indicate a temperature similar to that of the present 
Mediterranean (3), and to Spitsbergen and the Arctic Ocean. 
The remainder of the warm Tethys current travelled on 
between the Americas, and finally emerged again into the 
Pacific Ocean. Evidently there is no difficulty in accounting 
for the corals of the Tethys and Volga Seas or the rich 
vegetation of the valleys opening off them. 

In the Southern Ocean also, between the horns of the great 
crescent of Gondwanaland, there is no reason to suppose 
that the surface temperatures differed greatly from those 
prevailing at present in the same latitudes of the South Indian 
Ocean. The absence of the great masses of floating ice 
derived from the Antarctic would have tended to raise the 
temperature of the whole ocean, but, on the other hand, the 
Southern Ocean was mainly limited in the north by land in the 
neighbourhood of 40 S. instead of extending to the equator, 
and the absence of the supply of equatorial warm water 
would tend to balance the absence of the supply of ice. We 
have also to take into account the volcanic dust veil. At 
present the mean temperature of the ocean surfaces in latitude 
40-50 J is about 50 F., and we shall probably not be far out 
if we take the same figure for the temperature of the surface 
water south of Gondwanaland at the beginning of the Upper 
Carboniferous glacial period. When the land ice reached the 
sea over wide fronts, the temperature of the surface must 
have fallen much lower. 

One other point about the distribution of the seas is worthy 
of notice, namely, the long gulf extending from the Arctic 
regions into the heart of Eastern America, that is, into the 
only region outside Gondwanaland which appears to have 
been indubitably and severely glaciated during the Upper 

P. Lasareff (see Chapter III.) made an experimental 
reconstruction (see page 78) of the ocean currents of the 



Middle Permian (Fig. 30). This agrees in general with 
Fig. 29, especially in the warm current through the Tethys 
Sea and the branch across the North Pole. The Gulf between 
Nordatlantis and Nearctis is not shown (this of course may be 
due to the time-difference) but instead there is a cold current 
from the pole running down the western coast of Nearctis, 

Fig. 30. Ocean currents of Middle Permian. After Lasareff. 

which would be equally effective in giving a more severe 
climate to the present North America than that of Europe. 

We have now to discuss the system of the winds over 
Gondwanaland and the neighbouring seas. The " planetary " 
circulation of the atmosphere (Chapter II.) requires a belt 
of low pressure over the equatorial regions, a series of 


anticyclones in about latitudes 30 north and south, followed 
by belts of low pressure and storms in temperate or sub-polar 
latitudes. This system is modified by the land and sea 
distribution, which gives a tendency for high pressure over the 
land in winter and over the sea in summer. The geographical 
disturbance of the planetary circulation is now so great over 
Asia that the anticyclone does not develop there at a)l in 
summer, while in winter it attains a great intensity and is 
displaced some distance north of 30 N. The deflection of the 
entire Equatorial Current northwards into the Tethys Sea 
would probably suffice to maintain the temperature of middle 
latitudes north of Gondwanaland permanently above that of 
middle latitudes south of Gondwanaland, introducing an 
effect of permanent summer in the Northern Hemisphere and 
permanent winter in the Southern Hemisphere. We should 
expect to find a permanent low-pressure area over the Tethys 
Sea and the Volga Sea, while the normal sub-tropical anti- 
cyclone was developed oft the southern coast of Gondwanaland. 

Under these conditions, with an area of high pressure 
to the south of the equator separated by a very long and 
lofty continent from an area of low pressure north of the 
equator, what would be the wind system over the plateau ? 
We have no close parallel at present for guidance ; the 
nearest approach is found over Asia and the Indian Ocean 
during the south-west monsoon, but most of India is now 
at a comparatively low level, forming a plain out of which 
the Himalayan ridge rises steeply to a great height. The 
highest temperature and lowest pressure are found near 
Jacobabad to the south of this ridge. A large portion of the 
air which enters India from the high pressure area to the south 
is accordingly deflected to flow westward parallel with the 
mountain barrier ; it is only in the north-east that the air 
crosses at least the Khasi Hills (giving the enormous rainfall 
of Cherrapunji) and possibly the main ridge of the Himalayas. 
If the area of lowest pressure were to the north of the latter 
range, would the stream lines run directly across it ? 

The power of air currents to cross high ridges of land 
probably depends to a large extent on the steepness of the 
slope. The west-south-west winds from the Arabian Sea 
are able to cross the Western Ghats, for the greater part of their 
length more than 4,000 feet in height (4), and descend on the 


other side as a dry wind. The obstacle presented by the 
Himalayas, exceeding 12,000 feet, may be due quite as much 
to the steepness of their southern slopes as to their great 
height. If the ground sloped more gradually from Southern 
India or beyond to the Tibetan plateau, it seems probable 
that the height would be a less serious obstacle. 

Let us now suppose that the discussion of this first problem 
has shown that the air, starting from the Southern Ocean as 
a powerful south-east trade wind and changing to south-west 
as it crosses the equator, will climb steadily up the surface of 
the high ground, and, crossing the crest, will descend its 
northern slope. Under these conditions, what would be the 
general temperature and weather over the area ? The air 
starts at a temperature of about 50 F. and a humidity 
approaching saturation ; if the temperature of the plateau 
surface is mainly above 50 F., the air will not part with its 
moisture readily and the cloud amount will be small, but if 
the temperature of the surface is below 50 F., the sky will be 
mainly overcast and the precipitation heavy. 

Here again much seems to depend on the topography. 
If the southern margin of the continent was formed by a 
wide plain at a low level, the air would not at once be forced 
to rise sufficiently to develop an extensive cloud layer, the 
low land surface would therefore be exposed to intense 
insolation and would be very hot, and the air would be 
warmed up by contact with it to such an extent that it might 
rise to very high levels before it again became saturated. If, 
on the other hand, the ground rose fairly steeply from the 
sea to an elevation of 2,000 feet or so, the formation of 
a thick cloud layer would begin before the air had time to 
warm up. 

In North-eastern India, where the air crosses the Khasi 
Hills and enters at least the foot-hills of the Himalayas, the 
cloudiness during the south-west monsoon is very great. 
The mean cloudiness (8 a.m.) at Cherrapunji, Darjiling, 
and Mercara during June, July, and August is nine-tenths 
of the sky, while the relative humidity is 95 per cent. Let 
us suppose that, the southern coast of Gondwanaland being 
sufficiently steep to raise the incoming trade wind above 
the saturation level, the mean cloudiness was nine-tenths. 

The temperature of the air at any point is governed by 


the quantity of heat which it originally contained, plus the 
heat which it gains mainly from the surface, minus the heat 
which it loses, mainly by radiation. In the conditions 
postulated, the cloud would probably be rather thick ; let 
us assume that it had a mean density of 3. Then according 
to the measurements of B. Haurwitz (see Chapter VIL) the 
amount of radiation penetrating a cloud cover of nine-tenths 
would be barely half that received with a cloudless sky. This 
is less than the average amount received at present in latitude 
4O-50. If we take into account the loss by scattering, 
especially great during the Upper Carboniferous because 
of the great amount of volcanic dust, which would be more 
or less proportional to the total solar radiation and therefore 
greater in low than in high latitudes, we see that the amount 
of solar radiation available for warming the earth's surface 
was probably less over Gondwanaland than over the ocean 
to the south of it. 

To this it might be objected that the under surface of 
a cloud layer is also effective in reflecting the terrestrial 
radiation back to the surface of the earth, and that this would 
redress the balance. But it was shown in Chapter VII. 
that the reflecting power of clouds for long-wave terrestrial 
radiation is much less than their reflecting power for short-wave 
solar radiation. 

From this it appears that if the air entering Gondwanaland 
from the south immediately formed a cloud layer, it would 
not gain any heat from the surface of the plateau. If it 
continued to rise along a sloping surface, it would continue 
to cool by expansion and form fresh cloud. Taking the 
initial temperature of the air as 50 F. and the vertical tem- 
perature gradient as 3 F. per 1,000 feet, the snow-line would 
be reached at a height of 6,000 feet. There is no evidence 
against the supposition that at the beginning of glaciation a 
large part of Gondwanaland was above this height, so that 
under the conditions postulated, extensive snowfields could 
develop. The supply of snow would be ample, since preci- 
pitation would go on throughout the year, instead of for a 
few months only as in India, although, of course, the monthly 
totals would not equal those recorded at the wettest stations 
during the height of the monsoon. Hence there would be 
a plentiful supply of ice to extend below the snow-line 


on the southern slopes, and to reach the sea along broad 

It may be remarked here that owing to the absence of 
strong seasonal contrasts, the formation of ice-sheets would 
probably be very susceptible to slight changes of temperature 
and snowfall. At present glaciers form on high mountains 
near the equator, but owing to the steepness of the mountain 
slopes they descend rapidly to warmer levels and melt. Given 
larger gathering grounds, suitable high-level basins in which 
the ice could accumulate, and weakened solar radiation to 
lessen melting, the rapid development of glaciers into ice- 
sheets would appear to be inevitable. Once the snow cover 
had been formed and inland ice-sheets had begun to develop, 
conditions would at first be very favourable for their rapid 
growth. The ice surface would reflect a considerable part of 
the weakened radiation which penetrated the cloud, so that 
the surface would be very cold. On the southern slopes, 
south of the equator, there would be a tendency for the air 
drainage to form a north-west wind, which would come into 
opposition with the south-east monsoon-trade wind. At 
the surface both winds would probably prevail in turn first 
the relatively mild and moist south-east wind with a light 
snowfall, then an interval of calm, followed by a blizzard 
from the north-west, lifting the moist air of the south-east 
wind bodily and bringing a burst of heavy snow. But these 
changes would be limited to a shallow surface layer, and over 
all the south-east wind would blow steadily and the skies 
would be heavily clouded. The ice-sheet would merely add 
to the effective height of the land. 

But we can go much further than this. Once the ice-sheet 
had been formed, it would by its own cooling power and by 
the reflection of solar radiation from its surface back to space, 
effectively lower the snow-line over the glaciated area and its 
immediate neighbourhood, and thus enable the ice to spread 
over low ground previously unglaciated, or to survive a sub- 
sidence of the ground on which it rested below the original 
snow-line of 6,000 feet. Once a large ice-sheet has been formed, 
its persistence is probably almost independent of the latitude, 
and it can only be destroyed by the cessation of the supply 
of snow, by a great increase of the ablation, or by a subsidence 
of its bed below sea-level. Hence, during the later stages 


of the Upper Carboniferous glaciation, the land surface, 
worn down by glacial erosion and depressed by the weight 
of the ice, may actually have been a low plain, instead of 
the lofty plateau which we have supposed necessary for its 
inception. So long as the temperature difference was 
maintained between the Tethys Sea and the Southern 
Ocean, the snowfall would have been sufficient and the 
ablation small, so that the destruction of the Upper Carbon- 
iferous ice-sheets probably came about by subsidence and, 
in fact, the boulder clays are generally overlain by marine 

The greater part of the Gondwanaland ice-sheet was 
apparently formed on the southern slopes of the continent, 
where conditions were most favourable. There can have 
been little, if any, snowfall on the northern slope, and for a 
glacier to reach the Tethys Sea an exceptional topography 
would be required. In fact, it happened in only one region, 
India, and if we may judge from the thinness of the deposits, 
for a short time only. We may suppose that there the main 
watershed lay very far north, and that the high ground formed 
a sort of funnel through which a large amount of rising air 
was forced to pass. Somewhat similar conditions give rise 
to the abnormal rainfall of Cherrapunji at present. This 
high ground would form a very rich gathering ground for 
snowfields. If, then, a valley on the northern slope cut back 
deeply into this high ground, it might well receive sufficient 
ice for a glacier to reach the sea. 

The high temperature of the Tethys Sea and the Volga 
Sea would give a very favourable climate to Angaraland 
and the eastern parts of North Atlantis. The warm seas 
would keep the temperature high and the great evaporation 
would give rise to a heavy rainfall, making the low broad 
valleys among the mountains, open to the warm air from the 
sea, very favourable for the growth of a rich vegetation, 
able to give rise to the coal measures. These conditions 
would extend to the southern parts of Nearctis, where a 
small inlet seems to have existed in such a position that it 
carried the oceanic influences into the heart of the American 
coal district. The glaciers of Europe, if they existed at all, 
were probably not more than small mountain glaciers, such 
as can develop in any latitude provided the mountains are 


sufficiently high and the precipitation sufficiently heavy, 
and we know that in Late Carboniferous Europe there were 
both mountain ranges and heavy precipitation. The more 
extensive glaciation of North America can be associated with 
conditions in the Arctic Ocean. 

We have seen that the Volga Sea was occupied by a warm 
current which carried mild conditions as far as Northern 
Russia, while even in Spitsbergen there was sufficient 
vegetation to form workable coal beds. Evidently, where 
the Volga Current entered the Arctic Ocean, there was an 
ice-free area similar to that now formed by the Gulf Stream 
Drift, but larger. The Volga Current was more powerful 
than the Gulf Stream Drift, as we should expect it to be 
from the more favourable topography. Conditions between 
Angaraland and Nearctis are not sufficiently known to decide 
whether a similar warm current entered the Arctic between 
these continents, or whether they joined or approached so 
closely as at present as to prevent any warm current 
from passing between them. The map of this region is 
rather hypothetical, but the climatic conditions suggest 
that there was no such current. The North American 
glaciation seems to require the presence of a floating cap 
of sea ice to the north of Nearctis. In Chapter I., however, 
we saw that at present the Arctic Ocean is only a few degrees 
below the critical temperature required for the formation 
of such an ice-cap. At present, heat is carried into the 
Arctic by the Gulf Stream Drift and by the powerful warm 
south-west winds associated with the Icelandic minimum 
and its north-easterly extension. In the Upper Carboniferous 
the Volga Current probably supplied more heat than the 
Gulf Stream Drift, but the barometric distribution was probably 
less favourable that at present for the warm south-west winds, 
since the alignment of the Volgan cyclone was probably 
west to east instead of south-west to north-east. The im- 
portance of the latter point for the Quaternary glaciation 
has been well brought out by the late F. W. Harmer (5). 
We may regard the balance as about even, but the addition 
of a Bering Sea Current would other things being the same 
certainly raise the Arctic temperature above the critical 
point and bring about an ice-free Arctic Ocean. Hence 
it seems probable that there was little or no flow of warm 


water into the Arctic Ocean through a channel between 
Nearctis and Angaraland. 

Fig. 29 shows a long gulf extending from the Arctic far 
into the temperate zone between Nearctis and North Atlantis. 
This gulf, open to the Arctic but closed to the south, must 
have exerted a very powerful effect on the climate of the 
neighbouring land. It is doubtful if it froze in winter into 
a continuous surface of ice ; the influence of the warm sea 
to the south of the narrow land barrier may have been sufficient 
to keep it open, but if, as we suppose, the Arctic Ocean to 
the north was ice-covered, the waters of this gulf must have 
carried a great deal of loose floating ice in winter and spring. 
That this was so is shown by some of the North American 
glacial deposits, which contain large boulders apparently 
transported by icebergs or shore ice. Even in summer the 
water must have been very cold. There was probably a 
slow circulation, southward along the western side of the 
gulf and northward along the eastern side, similar to that 
now found in Baffin Bay, except that the latter being open 
to the south, the ice can escape into the Labrador Current. 
The close neighbourhood of this cold water to the north of the 
isthmus and the warm water of the Tethys Sea to the southward 
must have given rise to a tendency for great storminess, 
heavy rain, and dense fogs, weather similar to that now 
prevailing on the Newfoundland Banks. Given some mountain- 
ous country, such as we know to have been present, conditions 
here were very favourable for a moderate glaciation. 

Finally, we have to consider the climate of the Antarctic. 
According to Wright and Priestley (6), the Antarctic continent 
was not glaciated during the Upper Carboniferous ; instead, 
there was a dry and wind-swept plateau subject to severe 
frost action, and more favourable conditions in sheltered 
coastal lowlands in which a fairly rich flora was able to develop. 
Of course these conclusions refer only to the coastal parts of 
the continent ; the interior may have been glaciated to some 
extent, but even so the fact is surprising ; the general con- 
ditions over the Antarctic appear to have been no more 
favourable than at present, and we should have expected to 
find it glaciated. Probably it was not that the temperature 
was too high, but that the snowfall was too low compared 
with the ablation. With regard to present conditions, it is 


not certain that if the Antarctic ice-sheet could be melted 
entirely away, it would be re-established under the present 
climatic regime ; it may be simply a survival from the 
Quaternary Ice-Age. It is quite possible that if the greater 
part of the surface of the Antarctic during the Upper 
Carboniferous consisted of a fairly level plateau, the climatic 
conditions would resemble those of Northern Siberia, the 
very cold but comparatively dry winter not giving enough 
snowfall to persist through the summer. 

Thus we see that starting from a restoration of the dis- 
tribution of land and sea during the Upper Carboniferous 
on the basis of the existing positions of the continental massifs, 
and deducing the system of winds and ocean currents and the 
local climatic conditions in accordance with meteorological 
experience as represented by the nearest modern analogies, 
we arrive at a very fair reconstruction of the peculiar 
climatology of this period. The critical assumption is the 
considerable elevation of the central parts of Gondwanaland, 
but this is not entirely unsupported by evidence, and is at 
worst not less hazardous than the extensive migrations of the 
continents through some fifty degrees of latitude. We find 
that outside Gondwanaland the only extensive glaciation 
occurred in Nearctis exactly where we should expect it, 
while on the theory of continental drift this region would lie 
on the equator and would be unglaciated. The main difficulty, 
the non-glaciation of the Antarctic, is common to both theories. 

There is one peculiarity in the action of volcanic dust 
which deserves mention. In any latitude its cooling power 
is about proportional to the solar radiation received at the 
limit of the atmosphere, while its warming power, though 
much smaller, is proportional to the radiation from the 
earth's surface and the lower layers of the atmosphere. The 
cooling varies regularly with latitude, being greatest at the 
equator and least at the poles, while the warming is greatest 
in the warmest regions. Hence the effect is to increase the 
abnormalities of temperature brought about by the dis- 
tribution of land and sea ; Gondwanaland was cooled much 
more effectively than the Tethys, and in Spitsbergen during 
the polar night the effect was pure gain. 

Summing up, it appears that extensive glaciation in low 
latitudes required at least three, and possibly four conditions : 


1. The diversion of the whole of the equatorial ocean 
current into the Northern Hemisphere, which thereby 
became abnormally warm. 

2. An extensive elevated continent along the equator, 
but extending much farther into southern than into 
northern latitudes. 

3. A southern ocean shut off by land barriers from all 
warm currents. 

4. Possibly, a general refrigeration which might be due 
to the presence of abnormally large quantities of 
volcanic dust. 

So far as I can discover, these conditions occurred only 
once in geological time, and that occasion coincided with 
the only occurrence of extensive glaciation in low latitudes. 
Further, the climates of other parts of the world are such 
as would be expected from them. The Coal Measures 
become, not a violent negation of the possibility of glaciation, 
but a necessary complement to it. It seems to me that this 
geographical explanation is simple and natural, and does not 
violate probabilities as does the arbitrary shifting of continents 
and poles. 

In the same year (1926) that I first published this geo- 
graphical hypothesis of the Carboniferous glaciation, a some- 
what similar theory was put forward by C. Schuchert (7) 
who believed, however, that the date of the glaciation was 
definitely Permian. He writes : c< It is in the youthful 
topography, the enlarged continents and the peculiar con- 
nexions of the lands that seemingly are to be sought the 
reasons for the Permian Ice-Age. . . . This holding in of so 
much of the waters of the Antarctic Ocean, combined with 
the moist climates in the Southern Hemisphere and the 
general highland condition of much of the world in early 
Permian time, will be the explanation for the peculiar 
position of the continental ice-masses of the Southern 
Hemisphere." Later, Bailey Willis (8) attributed the glacia- 
tion of South America and South Africa to refrigeration 
of the South Atlantic, shut off from warm currents. The 
centres of glaciation were near the warm seas, which provided 
moisture by rising over wedges of colder air, giving cold 


foggy summer weather, favourable to a low snow-line. He 
considers that the glaciation of India was due to local high 
mountain ranges. 


(1) ARLDT, TH. " Handbuch der Palzeogeographie." Leipzig, 1919. 

(2) HANDLIRSCII, A. '* Die Bedeutung dcr fossilen Insekten fiir die Geologic." 

Wien, Mitt. Geol. Ges., 3, 1910, p. 503. 

(3) STAFF, H. v. " Zur Entwicklung der Fusuliniden." Zentralbl. f. Min. y 

Geol. und Pal., 1908, p. 699. 

(4) SIMPSON, G. G. " The south-west monsoon." London, Q,. J. R. Meteor. 

Soc., 47, 1921, p- I5 1 - 

(5) HARMER, the late F. W. " Further remarks on the meteorological conditions 

of the Pleistocene epoch." London, Q,. J. R. Meteor. Soc., 51, 1925, p. 247. 

(6) BRITISH (TERRA NOVA) ANTARCTIC EXPEDITION, 1910-1913. " Glaciology," 

by C. S, WRIGHT and R. E. PRIESTLEY. London, 1922. 

(7) SCHTJCIIERT, C. " The palseogeography of Permian time in relation to the 

geography of earlier and later periods.*' Proc. 2nd Pan-Pacific Set. Congr., 
1926, p. 1,079. 

(8) WILLIS, BAILEY. " Isthmian links." New York, Bull. Geol. Soc. Amer., 

43, 1932, p. 917- 



OF the four great ice-ages, the first two, the Lower 
Proterozoic, the Upper Proterozoic-Lower Cambrian, 
and the last, the Quaternary, were developed mainly 
in what are now temperate latitudes, while the third, the 
Upper Carboniferous, found its maximum extent in regions 
now not far from the equator. The Upper Proterozoic- 
Lower Cambrian glaciation was apparently similar in many 
respects to the Quaternary, but as yet we know so little about 
it that no detailed discussion is possible. In this chapter 
it is proposed to refer briefly to the meteorology of the 
Quaternary period. 

At the maximum of the Ice-Age, E. Antevs (i) estimates 
that the ice-sheets occupied an area of about 1 3 million square 
miles. Of these 4^ million were in North America, i J million 
in Europe, i^ million in Asia, and about 5 million in the 
Antarctic. The remainder was made up of the expanded 
ice-sheet of Greenland and relatively small areas in Australia, 
New Zealand and South America. The present ice-covered 
area of about 6 million square miles is almost entirely in the 
Antarctic and Greenland. It is not certain that all areas 
reached their maximum at about the same time, but two lines 
of evidence suggest that this is nearly true. The first is the 
lowering of sea-level by the abstraction of water, which 
according to Antevs's estimate would amount to 305 feet 
below the present if all the ice-sheets reached their maximum 
extent and thickness together. Various estimates have been 
made from the present depths of shore deposits, coral reefs, 
etc., which give a minimum figure of about 260 feet. The 
second line of evidence is the depression of the snow-line in 
North, Central and South America during the latest (Wurm) 
glaciation, which has been remarked on by many authors, 
and which is almost constant right across the equator, 



increasing somewhat in regions of heavy rainfall and decreasing 
in dry regions : 

Latitude .... 45 N. iyN. 10 N.-20 S. 408. 
Depression of snow- 
line (feet) . . . 2,300 3,000 1,300-2,000 3,300 

The character of the ice-sheets and glaciers varied. In 
Northern Europe, North America and presumably also in 
Greenland and the Antarctic they were several thousand feet 
thick and spread out actively from various centres. Siberia 
was for long considered to have had only minor mountain 
glaciers, but recent work summarised, e.g., by R. F. Flint and 
H. G. Dorsey (2) shows that there was at one time a large 
ice-sheet in north-west Siberia extending over the Arctic 
shelf, though the ice was thin and inactive. Farther east 
and south there were extensive piedmont glaciers. In South 
America the glaciers were also mainly of the piedmont type. 

The ice-age was divided into glacial and interglacial periods 
by a series of large-scale advances and recessions of the ice. 
These are best known from Europe and North America, 
and appear to run closely parallel in the two continents. 
The succession is summarised by F. E. Zeuner (3), K. 
Bryan (4) and others as follows (the youngest at the top) : 

Alps. N. German Plain. Continental U.S.A. 

Wurm Weichsel (including Wisconsin (including 

Warthe) lowan) 

Riss Saale Illinoian 

Mindel Elster Kansan 

Gunz ? Nebraskan (Jerseyan) 

The Mindel and Riss glaciations were the most extensive. 
The Mindel-Riss interglacial (Yarmouth Interglacial in 
America) was very long, of the order of 240,000 years, and was 
generally mild, but was interrupted by at least one colder 
period which did not reach glacial intensity. The Riss 
glaciation had a double maximum in Europe at least. The 
Riss- Wurm interglacial was short and mild but was inter- 
rupted near the middle by a period of sub-arctic conditions 
in Jutland. 


The Wurm glaciation was less extensive than the Riss ; 
it comprised three maxima, each of less intensity than the 
preceding one, separated by sub-arctic or even cold-temperate 
conditions. The recession after the third peak was inter- 
rupted by several halts or slight readvances. 

In other parts of the world the succession is less complete. 
In Kashmir and neighbouring territories F. Loewe (5) 
recognises three glaciations, which he correlates with Mindel, 
Riss and Wurm, decreasing in intensity, and there are probably 
traces of the same three in East Africa ; in both cases, however, 
the correlation is not certain. In nearly all other parts of the 
world the remains of only two glaciations are found, pre- 
sumably representing the Riss and Wurm, the former always 
being much the more extensive. In Siberia for example there 
was no Wurm ice-sheet, only mountain and piedmont glaciers. 
It is not yet certain that the earlier glaciations did not occur ; 
their moraines may have been destroyed by later advances. 

In discussing the cause of the Quaternary Ice- Age, it is 
necessary to distinguish between the Ice-Age as a whole, 
and the succession of glacial and interglacial periods. In 
Chapter XII. it was shown that the most probable cause for 
Ice-Ages was elevation and mountain building in extensive 
high continents, limited accession of warm ocean currents to 
high latitudes, and probably much volcanic dust in the 
atmosphere, all of which factors were present at the beginning 
of the Quaternary. A decrease in the amount of CO 2 in the 
atmosphere (Chapter VI.) may have been a contributory 
factor. With the exception of volcanic dust, however, these 
are all stable factors, which would not change sufficiently 
rapidly to account for the succession of glacial and interglacial 
periods. For the latter therefore some other explanation 
must be found. 

The fact that the last two glaciations at least began and 
ended more or less together in all parts of the world is highly 
significant. In a minor degree it is paralleled by the recession 
of the glaciers everywhere within the past hundred years. It 
shows that glaciations in different regions do not depend only 
on local conditions, but are mainly controlled by some world- 
wide factor such as the temperature of the oceans, the heat 
received from the sun, or the circulation of the atmosphere, 
or by some combination of them. The late beginning of 


glaciation in the tropics and Southern Hemisphere, if confirmed, 
suggests that ocean temperature may be important, because 
the lag in cooling the oceans would be greater the farther 
removed the region is from the sources of cold water in the 
Arctic and North Atlantic. 

The first question concerns the lag between the occurrence 
of mountain formation and the beginning of glaciation, which 
was discussed in Chapter X. There are four possible reasons : 

1. The slow cooling of the oceans. 

2. The erosion of the mountains. 

3. The occurrence of a period of explosive volcanic activity. 

4. The occurrence of favourable astronomical conditions. 

An important factor in fixing the actual beginning of the 
Quaternary glaciation over the land must have been the general 
temperature of the sea. At the close of a long warm period 
the sea is warm throughout its whole depth ; there is none 
of the very cold bottom water which exists at present. This 
must be so, for the temperature of the sea depths cannot long 
remain lower than the temperature of the coldest part of the 
surface. Now the beginning of the Quaternary glaciation 
was a period of great elevation in most parts of the north 
temperate belt. The gap between Greenland and Norway, 
which at present conducts the Gulf Stream into the Arctic 
Ocean, was greatly narrowed if not completely closed. Bering 
Strait probably differed little from its present condition, and 
there may have been an open channel to the west of Greenland. 
Now there are some interesting peculiarities in the develop- 
ment of the Quaternary glaciation which may have a bearing 
on this question of the cooling of the seas. The first glaciation 
of Europe was most extensively developed in Scandinavia 
and North Russia ; the British Isles were probably not 
glaciated until later. The corresponding glaciation in America 
was developed in the Rocky Mountains of British Columbia 
and in Labrador, but not in the central parts. The glaciation 
of British Columbia was apparently an enormous development 
of valley and piedmont glaciation due to the great height of 
the mountains, but the North European and Labradorean 
centres developed true inland ice-sheets. If we suppose that 
the elevation of the Wyville Thomson ridge between Greenland 


and Scotland above its present level shut out the Gulf Stream 
from the coast of Norway, the Arctic Ocean would lose almost 
all the supply of heat formerly carried into it by ocean currents 
and its temperature would begin to fall. The ocean south 
of the Wyville Thomson ridge would still be very warm, 
however, arid the winds must have brought a considerable 
amount of heat across the land barrier. It is difficult to 
estimate the time which would be required under these 
conditions for the thorough cooling of the Arctic Ocean. 
Most of the ice formed in the Arctic at present begins with the 
freezing of a surface layer of relatively fresh water brought 
down by the great rivers which enter the basin, and which, 
owing to its smaller density, floats on the main mass of warmer 
but more saline water. Probably this water must freeze 
fairly near the coast, otherwise the storm winds would break 
it up and mix it with the underlying salt water. By analogy 
with what happens at present at the junction of the Labrador 
Current and the Gulf Stream, we can say that the fresh layer 
would become salt more quickly than it would warm up (6). 
The resulting mixture would be heavier than both the upper 
and lower layers, and would therefore sink. But while the 
main oceans were still warm, it seems probable that the heat 
transferred by southerly winds would suffice to keep this layer 
of fresh water liquid long enough for the mixing process to 
destroy it. Thus we conclude that the formation of a cover of 
floating ice probably did not follow immediately on the 
elevation of the Wyville Thomson ridge, but had to wait until 
the cooling of the main oceans had progressed some way. 

At first sight it might seem that the accumulation of cold 
bottom water could not possibly affect the atmospheric 
processes which go on above the surface of the oceans. Such 
an influence does take place, however, especially off the western 
coasts of the continents, where cold bottom water wells up to 
replace the surface water driven away by easterly winds. 
Investigations into the effect of the Trade winds on the surface 
temperature of the North Atlantic have shown that the 
North-east Trade, blowing off the coast of West Africa, does 
actually bring up a large amount of cold water from the 
underlying layers. This cold water has the effect of lowering 
the temperature of the Gulf Stream, and ultimately the 
surface temperature of the North Atlantic between the United 


States and Ireland, probably by several degrees. If the 
depths of the oceans were much warmer than at present, 
this cooling influence would not exist. The meteorological 
effects of upwelling cold water on the western coasts of South 
America, South Africa, and Australia are extraordinarily 
marked, being largely responsible for the desert character of 
those coasts. It is probable, however, that this cold water 
comes, not from the greatest depths, but from some inter- 
mediate layer, and that a certain accumulation of cold water 
could take place without affecting surface conditions. 

We have no means of knowing how long it took to cool 
the main body of the oceans, but it was certainly a very long 
time. As an example of the quantities involved, if we suppose 
that all the thaw water of the ice, both land and sea ice, which 
melts each summer in both hemispheres, sank to the bottom 
of the oceans and spread out there, it would take between 
ten and twenty thousand years to fill the oceans with cold water. 
Immediately after the formation of the Wyville Thomson 
ridge, the annual supply of cooled water was probably not 
so great as the present annual melting of ice, and at first it 
was not ice-cold. When we take into account also the cold 
water which wells up in the tropics and becomes warmed 
there, so that it has to be cooled again, we see that the thorough 
cooling of the oceans must have taken several times, perhaps 
many times, ten thousand years. It is unlikely, however, that 
this effect could have caused a lag of millions of years. 

The second stage in the oncoming of the Quaternary 
Ice-Age would occur when the general temperature of the 
oceans had fallen low enough for a covering of floating ice 
to develop over the Arctic Ocean. This would result in a 
great cooling of the lands washed by that ocean Greenland, 
Norway, and Northern Russia. The Labrador Current may 
have been in existence before, but now it would carry great 
quantities of floating ice, and there would be a great lowering 
of temperature in Labrador and Newfoundland. The decrease 
in the summer temperature would be greater than the decrease 
in the winter temperature, and there would also be a marked 
increase in the storminess and snowfall. All these regions 
would develop glaciers, which would speedily become ice- 
sheets. The glaciation of the mountains of British Columbia 
may have commenced earlier, but probably increased rapidly 


about this time, while the ice-sheet of the Antarctic probably 
reached the sea. This may have been the first or Gunzian 

The second cause of lag is the reaction of the elevated land 
areas to erosion. The action of frost and running water 
removed great quantities of rock, much of which found its 
way into the sea. The lightening of the load caused further 
uplift, but the topography now being irregular, the higher 
peaks were at a greater height than before, while the valley- 
heads formed suitable gathering grounds for the accumulation 
of snow drifts. It is in such hollows that snow dritts persist 
longest in Scotland. 

Finally, when all other factors were favourable, it is possible 
that either a period of plentiful volcanic dust or a period of 
decreased radiation in summer due to astronomical causes, 
by keeping down the summer temperature, was the actual 
immediate cause of the beginning of glaciation. 

Once the ice-sheets had formed, by raising the effective 
height still further (both by adding ice to the land and sub- 
tracting water from the sea), by reflecting solar radiation 
and cooling the area around them, and by shedding ice into 
the sea and so depressing ocean temperatures still further, 
they would tend to maintain themselves and spread, until for 
some reason they became unstable. We must now examine 
the possible causes of the break-up of the ice-sheets. These 
are : 

1. A lowering of the level of the land. 

2. A general rise of temperature. 

3. A decrease in snowfall. 

Since ice weighs about one-third as much as the average 
rock, the accumulation of 3,000 feet of ice is equivalent to 
adding the weight of 1,000 feet of rock to the land. This 
additional weight gradually depressed the land surface, 
though with a considerable lag, and brought the margins of 
the ice-sheets under the action of the sea, causing for example, 
floating ice-barriers which broke away as icebergs. The 
area of the ice-sheets and consequently their cooling power 
diminished, initiating an amelioration of climate. The 
process might be carried far enough for the ice to disappear 


more or less completely. After the load was removed the 
land would begin to rise again, causing a return of glaciation. 
This process might account for the division of a glacial period 
into two or three peaks, and possibly, though this is more 
doubtful, for the shorter interglacial periods. Moreover, 
since each glaciation would wear down the high ground and 
deposit the material round it in the form of moraines, we 
should expect each glacial recurrence to be less severe than the 
preceding one until the topography became unsuitable for 
glaciation. It cannot account for the long Mindel-Riss 
interglacial, but during the latter there was a great deal of 
earth-movement and volcanism in many parts of the world, 
which would eventually cause a return of glaciation. It is 
known that after the Wurm glaciation the land in the centres 
of greatest ice accumulation continued to sink and in late 
Glacial time the central shores of the Gulf of Bothnia were 
depressed about 900 feet below their present level ; the 
recovery is still in progress. Similar subsidence and recovery 
should have followed each glacial advance but the evidence 
for interglacial oscillations of the same type was swept away 
by subsequent advances of the ice. Also, the oscillations of 
level may have been superposed on a steady sinking of the 
land, so that each rise was less than the preceding fall. This 
would account for the gradual decrease in intensity of 
successive glacial peaks. 

A world-wide rise of temperature could be due to the 
cessation of volcanic activity, to an increase of solar radiation 
(Chapter IV.) or to astronomical causes (Chapter V.). The 
first two are rather speculative ; moreover it is very doubtful 
whether the slight increase of radiation in middle and high 
latitudes which would result from a cessation of volcanic 
activity would have much effect on a full-grown ice-sheet. 
Astronomical effects also seem rather slight, but as the effect 
of increased warmth in summer would be reinforced by the 
greater cold of winter which would probably result in a 
decrease of snowfall, they cannot be ruled out. The good 
accord between Milankovitch's astronomical scheme and the 
succession worked out by F. E. Zeuner (Chapter V.) supports 
the idea that these small astronomical causes may actually 
have been the controlling factor in the glaciation of the 
Northern Hemisphere. These large ice-sheets would exercise 


a dominant effect on the ocean temperatures and atmospheric 
circulation, and so might well control the glaciation of other 
parts of the world. There is good reason to believe that the 
Pluvial periods of low latitudes were in fact controlled by the 
atmospheric circulation. In this connexion it is interesting 
to note that according to H. Mortensen (7) there was no 
pluvial period in the coastal desert of northern Chile. This 
desert exists because of the upwelling cold water of the 
Humboldt Current off the coast, which in turn is due to 
the south-east trade winds blowing off the coast. A strength- 
ening of these trade winds would therefore maintain the 
desert conditions. 

We come finally to the question of precipitation. The 
supply of precipitation would of course follow variations 
of solar radiation (Chapter IV.) but it is now generally 
recognised that the development of ice-sheets would itself 
cause changes in their supply of moisture. This was first 
suggested by V. Paschinger (8). 

In Chapter IX. we saw that in mountainous country, as 
we go upwards the total amount of precipitation increases 
to a certain level, above which it again decreases. With 
increasing height, also, the proportion of total precipitation 
which falls as snow becomes steadily greater. Hence we can 
distinguish a level of maximum rainfall, and above that a 
level of maximum snowfall. The latter is often very sharply 
marked ; it depends on the winter conditions, especially 
the general winter temperature of the lowlands, the vertical 
temperature gradient, and the relative humidity. The 
snow-line, on the other hand, depends mainly on the summer 
temperature. At present in the Alps the snow-line is about 
2,000 feet above the level of maximum snowfall. Suppose 
now the summer temperature decreases while the winter 
temperature remains unchanged. The snow-line will descend, 
and if the decrease of summer temperature reaches 6 F., the 
snow-line will coincide with the level of maximum snow-fall. 
The supply of snow available for glaciers will now be greatly 
increased, and this stage will see a great development of 
glaciers. Even if the cooling is uniform throughout the year, 
the snow-line will descend more rapidly than the zone of 
greatest snowfall. 

At present in polar regions the snow-line is below the 


zone of maximum snowfall, and these regions are widely 
glaciated. In the Tertiary period, the snow-line must have 
been above the snowfall maximum even in polar regions. 
Paschinger considers that the cooling of the temperate regions 
spread out from the poles, probably in the form of repeated 
cold waves (i.e., outbreaks of the polar front). Owing to the 
conservation of heat in the oceans, whence most of the moisture 
is evaporated, the total precipitation is not diminished at first, 
while the proportion which falls as snow is increased. Glaciers 
spread until they reach the sea or some warm lowland where 
ablation is rapid. Then as the seas cool, the snowfall 
diminishes, while the lowering of temperature due to the ice 
itself depresses the zone of maximum snowfall. At the same 
time, the development of glacial anticyclones cuts off the 
supply of snow in the interior, so that the snow-line rises, until 
it is again above the zone of greatest snowfall. The ice-sheets 
and glaciers now retreat. When the retreat has proceeded far 
enough, the secondary cooling due to the ice ceases to be 
effective, the level of maximum snowfall rises to the snow-line 
again, and the whole process recommences. This is 
Paschinger's conception of the meteorological cycle of a glacial 
period ; granted an initial cause, such as elevation, glacial and 
interglacial (or " intraglacial ") stages will repeat themselves 
regularly until the immense denudation effected by the ice 
lowers the mountains or at least the corries and depressions 
where snow can gather below the snow-line. He thinks that 
this stage has not yet been reached in Europe and that another 
glaciation is to be expected in due course. 

Paschinger points out that the relationship between the 
level of maximum snowfall and the snow-line accounts for 
many peculiarities of the Quaternary Ice-Age. In the 
continental mountain regions of Asia, with very cold winters 
and hot summers, the two levels are many thousand feet 
apart, and th$ glacial cooling was not, as a rule, sufficient to 
bring the snow-line down to the zone of heavy snowfall. 
Hence the development of glaciers and ice-sheets was less 
extensive than in Europe or North America. In equatorial 
regions, on the other hand, while both snow-line and maximum 
snowfall are at a great height, the former lies only a short 
distance above the latter, owing to the absence of seasons. 
A comparatively slight increase in the snowfall would bring 


them together, and cause a considerable extension of the 
mountain glaciers. 

This view of the sequence of events in an ice-age 
undoubtedly contains many elements of truth, and may 
well account to some extent for the alternation of glacial 
stages with what I have termed above " intraglacial " stages. 
The Mindel-Riss interglacial stands in a different category, and 
cannot be accounted for on any purely meteorological cycle ; 
it necessarily involves a cessation or great weakening and a 
subsequent renewal of the ice-forming factors. 

As was pointed out in Chapter II., the development of 
ice-sheets would cause changes in the atmospheric circulation 
and tracks of depressions, which would react on the supply 
of moisture. Besides the work of Flint and Dorsey, referred 
to in that chapter, there have been several other studies of 
American glaciation on these lines. Thus E. Antevs (9) 
considers that the Keewatin and Cordilleran ice-sheets in the 
west and centre developed first. The Keewatin and Scandi- 
navian ice-sheets caused a southward displacement of the 
Icelandic low which caused frequent north-east winds in 
Labrador. Once started, the Labrador ice-sheet was fed by 
cyclonic snowfall on its southern border. Ultimately the area 
of ice grew so large that the supply of snowfall in the central 
regions was insufficient to maintain it, and the ice-sheet began 
to decay. At this stage, however, depressions were still deflected 
southward, and Antevs thinks that the mountain glaciers and 
lakes south of the main ice mass may have reached their 
maximum during the earlier stages of the retreat. Later, 
however, the storm tracks shifted north again and brought 
about a rejuvenation of the ice-sheets and a repetition of the 
series of events. This process might account for short intra- 
glacial periods but not for recurrences after long interglacial 
periods as warm as or warmer than the present. 

There is no doubt that changes in the atmospheric circulation 
must have brought about changes in the centres of the ice-sheets 
and it is highly probable that their growth must eventually 
have resulted in starvation at the centre, but it seems unlikely 
that this would have resulted in their disappearance. It is 
also doubtful whether the North American sequence followed 
the lines of Antevs's argument ; K. Bryan (10) for example, 
states that American geologists believe in a progressive shift 



from east to west of the main ice-centre throughout the last 
(Wisconsin) glaciation, and R. F. Flint (n) thinks that the 
Labradorean and Keewatin areas were both parts of a single 
Laurentide ice-sheet fed by maritime air from the south and 
south-east and expanding southward and westward. 

Summing up, we find that for the occurrence of the ice-age 
as a whole the " geographical " theory seems to be the only 
adequate one, with possibly some help from CO 2 . The 
actual commencement of glaciation may, however, have been 
determined by some minor factor such as the astronomical 
situation or changes of solar radiation. The interglacial 
periods present the main difficulty, because of their close 
parallelism in different parts of the world. Astronomical 
causes seem to come nearest to filling the necessary conditions, 
but alternating depression and elevation, due to the accumu- 
lation and removal of the ice-load, are also probable, while 
cycles of solar radiation cannot be ruled out. The re- 
crudescence of glaciation after the Mindel-Riss Interglacial 
was due, at least in part, to renewed mountain building. 
Finally, the " intraglacial " oscillations were most probably 
caused by reactions between the ice-sheets and the circulation 
of the atmosphere. 

We must now briefly consider the climate outside the main 
areas of glaciation. 

The part of Central Europe sandwiched between the 
Scandinavian ice-sheet to the north and the Alpine glaciers 
to the south must have suffered from a severe climate, which 
has been studied by P. Kessler (12). He has three lines of 
evidence the climate in the neighbourhood of the present 
ice-sheets of Spitsbergen, Greenland, and the Antarctic, the 
flora and fauna, and the geological phenomena and all three 
present the same picture. The mean annual temperature is 
below freezing point, and although the summer may have a 
few short spells of warmth, the winters are very cold. On 
the margins of the Antarctic continent the summer climate 
is especially unpleasant. Although the temperature during 
a relatively warm summer month may average above freezing 
point, and may go as high as 40 or even 45 F. for a few 
hours, yet the persistently overcast sky, the frequent storms 
of snow and sleet, and the general unpleasantness of the 
weather, are worse than the cold of the interior. 


The conditions in these high latitudes, between ice-sheets 
and the sea, however, cannot be regarded as typical of those 
in Central Europe far from the Atlantic, especially if, as seems 
probable, there was a considerable area of land west of 

The study of the flora and fauna gives results of great 
interest. The similarity of the plants at high levels in the 
Alps to Arctic forms suggests that during the maximum 
extension of the ice these cold-loving species inhabited the 
low unglaciated ground north of the Alps, and after the ice-age 
they followed the retreating glaciers upwards to high levels. 
The general picture shows a region of tundra vegetation, 
inhabited by the reindeer, the woolly rhinoceros, and the 
mammoth. The geological phenomena earth-flows, block- 
trains, and mounds, ridges or terraces of angular and sub- 
angular material point to frost action on a huge scale, the 
earth and rocks moving down the valleys under the action of 
repeated freezing and thawing. The general climate of the 
region appears to have been highly abnormal ; the prevailing 
winds were probably dry glacial winds from the north-east, 
but these winds were shallow and were overlain at a small 
height by moist winds from the Atlantic, which sometimes 
descended to the level of the ground. The snow-line lay 
at 3,000 feet in the west and at 5,000 feet in the east, and in 
the hills the accumulations of snow carved out cirques or corries 
at these levels. These are mainly on the north-eastern side 
of the crests, and since Enquist has shown that the greatest 
accumulation of snowfall takes place on the lee-side, they 
indicate that the snow-bearing winds at a height of 3,000 to 
5,000 feet came from the south-west. 

The annual precipitation was small at low levels, but 
occasionally rain fell in torrential downpours. The evapora- 
tion was great, and one of the greatest peculiarities of the cold 
periglacial climate was that it could ape the formations ofthe 
hottest deserts. An important deposit was the loess, an 
accumulation of the finest wind-blown dust, and there were 
even small salt lakes in which layers of salt were formed. 
It is noteworthy that similar saline deposits are forming at 
present in restricted areas in Spitsbergen and Greenland, a 
fact which has some bearing on the evidence for Wegener's 
theory of polar movements. 


Outside the limits of the ice-sheets and of the peripheral 
zone of ice-winds, the weather was probably much as we know 
it to-day, but more stormy. This applies especially to the 
Mediterranean region, which must have had a heavy rainfall 
distributed more or less evenly throughout the year, instead 
of a moderate or scanty rainfall limited to the winter months 
as at present. These regions probably had the weather now 
found on the north-western coasts of Europe. Wandering 
storms penetrated into the Sahara, which was then one of the 
most genial regions on the globe, and this region, now a desert, 
appears to have been one of the main centres in which the 
human race rose to a dominant position in the world. 
H. v. Ficker (13) calculated that at the time of the maximum 
glaciation of the north-west Pamir the rainfall was four or 
five times as great as at present. 

The equatorial regions in general also had a greater rainfall 
than at present, though with local exceptions. Over the 
oceans the Trade winds, stronger in consequence of the 
greater temperature difference between the equatorial and 
polar regions, brought in more warm moist air than at present. 
The volume of air ascending in the equatorial belt of low 
pressure was therefore greater, and the rainfall in the Doldrums 
and over the eastern equatorial parts of the continents was 
heavier. The succession of glacial and interglacial periods in 
the northern continents was paralleled by a succession of 
pluvial and interpluvial periods in tropical Africa, and by 
advances and retreats of the mountain glaciers. The exact 
correlation is not yet determined, but may be as follows : 

A very early lake, the deposits of which have been described 
by E. J. Wayland (14) as Kafuan, may correspond with 
Gunz and Mindel. Wayland thinks it had two maxima 
separated by a period of earth movements. After a long 
dry interval a large lake (Lake Kamasia) formed from the 
junction of several existing lakes. E. Nilsson (15) calls 
this the Great Pluvial and equates it to the Riss. Lake 
Kamasia then dried up completely and the mountain glaciers 
disappeared. This interpluvial was followed by the Gamblian 
period of renewed lake-formation in each of the separate 
basins. Nilsson distinguishes four successive lake systems, 
the first three representing the three maxima of the Wurm 
and the fourth a late Glacial halt or re-advance. Between 


Lakes I. and II. and II. and III. there were lower lake levels, 
between III. and IV. the lakes dried completely. 

A similar succession can be traced over a large part of 
East Africa from the Nile Valley to Rhodesia though the 
stages of the last Pluvial have not been distinguished. The 
Upper Nile Valley, however, became desert early and K. S. 
Sandford (16) considers that in that region there were no 
changes sufficiently great to be called " Pluvial " and 
" Interpluvial." It is in fact likely that owing to the pre- 
ponderance of ice in the Northern Hemisphere the whole 
system of climatic belts was shifted southwards and that the 
increase of rainfall was much greater south than north of the 

The retreat of the ice-sheets shows a number of halts or 
re-advances marked by a series of terminal moraines. These 
present a similar appearance in North America and Europe. 
There were also a series of fluctuations of lake-levels in East 
Africa, which most probably represent the pluvial equivalents. 

The variations of lake levels do not necessarily represent 
very great changes of rainfall. In the Nakuru catchment 
area the present rainfall is about 37.} inches a year. R. E. 
Moreau (17) from botanical evidence considers that the 
average rainfall in the last of the Wurmian pluvial stages 
(Makalian) was about 44-50 inches, while during the arid 
Post-Makalian period, when the lakes dried completely, it 
cannot have been as low as 27 inches. 


(1) ANTEVS, E. " The last glaciation." New York, Amer. Geogr. Soc., 

Research Series, no. 17, 1928. 

(2) FLINT, R. F., and H. G. DORSEY. " lowan and Tazewell drifts and the 

North American ice-sheet." Amer. J. Sci., 243, 1945, p. 627. 

(3) ZEUNER, F. E. " The Pleistocene period ; its climate, chronology and 

faunal successions." London, Ray Soc., 1945. 

(4) BRYAN, K., and L. L. RAY. " Geologic antiquity of the Lindenmeier site 

in Colorado." Washington, Smithson. Misc. Coll., 99, no. 2, 1940. 

(5) LOEWE, F. " Die Eiszeit in Kaschmir, Baltistan und Ladakh." Berlin, 

%s. Ges. Erdkunde, 1924, p. 42. 

(6) SMITH, E. H. " The international ice patrol." Meteor. Mag., London, 

60, 1925, p. 229. 

(7) MORTENSEN, H. " Uber den Abfluss in abflusslosen Gebieten und das 

Klima der Eiszeit in der nordchilenischen Kordillera." Naturwiss> Berlin, 
16, 1929, p. 245. 

(8) PASCHINGER, V. " Die Eiszeit ein meteorologische Zyklus." s. Gletscherk., 

i3 1923, P- 29- 


(9) ANTEVS, E. " Correlation of Wisconsin glacial maxima." Amtr. J. Sci. t 
243A, 1945, p. i. 

(10) BRYAN, K., and R. C. GADY. "The Pleistocene climate of Bermuda." 

Amer. J. Sci., 27, 1934, p. 241. 

(11) FLINT, R. F. ** Growth of North American ice-sheet during the Wisconsin 

age." New York, Bull. geol. Soc. Amer., 54, 1943, p. 325. 

(12) KESSLER, P. " Das eiszeitliche Klima und seine geologischen Wirkungen 

im nicht vereisten Gebiet." Stuttgart, 1925. 

(13) FICKER, H. v. " Die eiszeitliche Vergletscherung der nordwestlichen 

Pamirgebiete." Berlin, SitzBer. Preuss. Akad. Wiss., 1933, 2, p. 61. 

(14) WAYLAND, E. J. " Rifts, rivers, rains and early man in Uganda." London, 

J. R. Anthrop. hist., 64, 1934, p. 333. 

(15) NILSSON, E. " Quaternary glaciations and pluvial lakes in British East 

Africa." Geogr. Ann., Stockholm, 13, 1931, p. 249. 

(16) SANDFORD, K. S., and W. J. ARKELL. " Palaeolithic man and the Nile 

valley in Nubia and Upper Egypt." Chicago Univ., Oriental Inst., 
PubL, vol. 17. Prehistoric survey of Egypt and Western Asia, Vol. 2, 
Chicago (1933). 

(17) MOREAU, R. E. " Pleistocene climatic changes and the distribution of 
life in East Africa." London, J. Ecol., 21, 1933, p. 415. 





IT is not many years since it was generally believed that 
variations of climate came to an end with the Quaternary 
Ice-Age, a period moreover which was placed hundreds 
of thousands of years ago. The post-glacial or " Recent " 
period was supposed to show merely a more or less rapid 
warming up to the present level, followed by a long period 
in which the climates of the different parts of the world were 
exactly as we now find them. It was the International 
Geological Congress at Stockholm in 1910 which first made the 
majority of geologists familiar with the existence of a warm 
period intercalated between the ice-age and the present. 
About the same time, a number of investigations in different 
countries combined to prove that the ice-age itself was not 
so remote as it had seemed to be, and that in fact the post- 
glacial " geology " of Europe was partly contemporaneous 
with the " history " of Egypt. But since the geological 
deposits undoubtedly point to changes of climate, slight indeed 
in comparison with the preceding ice-age, but still marked 
enough to leave their traces permanently written on the face 
of the earth, the unvarying climate of history is evidently a 
myth. The beginning of the " period of unchanging climate " 
has advanced later and later before the attacks of geologists, 
and now, in the minds of most of the authors who concern 
themselves with the subject, it apparently stands only a few 
centuries before Christ. But meanwhile a different, and more 
logical, view has arisen, namely, that the present does not 
differ from the past, that variations of climate are still in 
progress, which are similar in kind, though not in extent, 
to the climatic vicissitudes of the ice-age. 

There is, however, one point in which the " historical " 
period may be said to differ from the " geological " periods ; 
during the historical period the distribution of land and sea, 
the heights of the mountains, and the positions of the poles have 
changed only to a very slight extent. Hence we may regard 


the geographical factors of climate as practically constant 
during this period, and any climatic changes which we can 
discover and confirm must be attributed to non-geographical 
factors, and most probably to variations in solar radiation. 
Hence it is in the historical period that we are most likely to 
be able to trace the effect of solar radiation on climatic changes. 
Of course this difference between the " historical " and the 
" geological " periods is more apparent than real ; the 
length of the historical period is a few thousand years, while 
the length of even the subdivisions of the geological periods is 
to be expressed in hundreds of thousands or in millions of 
years. Nevertheless, we do seem to be living at present in a 
period of quietude relative to the Quaternary period ; the 
change from the Ancylus to the Litorina stages in the Baltic, 
for instance, represents a greater geographical variation than 
anything which has happened since. 

The interpretation of the term " historical period " adopted 
in this section is a somewhat liberal one ; it is essentially the 
period during which the vicissitudes of human life are known 
and dated to within a few centuries. Archaeologists are 
continually pushing back the boundaries of history, while 
astronomers, geologists, and others from time to time supply 
new fixed points or new chronologies. At the present time 
we have a more or less complete record of human history in 
South-western Asia since about 5200 B.C., and that date has 
been taken as the point of origin. For a study of the climatic 
changes during this period of 7,000 years, we have a variety of 
material. Instrumental records are of course of the greatest 
value, but reliable meteorological observations go back a 
mere three centuries, and for the greater part of the period we 
have to make the best of less direct evidence. The various 
lines of attack may be summed up as follows : 

1. Instrumental records and old weather journals. 

2. Literary records (accounts of floods, droughts, severe 
winters, and great storms). 

3. Traditions, such as that of the Deluge, which can 
sometimes be correlated with other data. 

4. Fluctuations of lakes and rivers, glaciers and other 
natural indices of climate, which can often be connected 
with historical events or dated by laminated clays. 


5. Arguments from the migrations of peoples, for which 
climatic reasons may be assigned with some show of 
probability. To this we may perhaps add the waxing 
and waning of civilisations. 

6. The rate of growth of trees, as shown by the annual rings 
of tree-growth, which can be correlated with the annual 

7. Geological evidence great advances or retreats of 
glaciers, growth of peat-bogs, succession of floras, etc., 
which can sometimes be dated approximately. 

The first and second sources of data, meteorological and 
literary records, and the seventh, geological evidence, are 
mainly exemplified in Europe, while the fourth and fifth 
sources provide the main mass of information for Asia and 
the fourth for Africa ; while the sixth, growth of trees, gives 
the only exact chronology for North America, where, however, 
it is highly developed. 

Instrumental meteorological records even in Europe date 
back for only about three centuries, in North America for 
two centuries, while in other continents they are practically 
confined to the last hundred and fifty years. Moreover, 
while old observations are of great interest in discussing 
variations of weather from one year to another, they are of 
less value in determining changes of climate extending over 
a long period. The accuracy of the early instruments is not 
always above reproach ; some of the early types of rain gauge, 
for example, do not make adequate provision against the 
re-evaporation of the fallen water. Defects of exposure may 
be a serious source of error. It was a common practice among 
early observers to expose their rain gauges on the roofs of 
houses, but gauges so exposed do not catch so much water as 
gauges exposed on the ground in open sites. Even when 
placed on the ground, they may have been too near to buildings 
or trees. Most of the sources of error tend to give a rainfall 
which is too small rather than too great, so that if the early 
instrumental records appear to indicate that the rainfall was 
smaller than at present, they must be regarded with suspicion 
unless they can be confirmed in some way. The rainfall 
minimum in England indicated by Symons in the eighteenth 
century was suspected for this reason ; the way in which it was 


confirmed is described in the next chapter, where, in addition, 
long rainfall records are discussed from other countries. 

There is a curious exception to the comparative modernity 
of instrumental meteorological records, namely, the measure- 
ments of rain in Palestine in the first century A.D. Hellmann ( i ) 
states that " the amount of rainfall then considered as normal 
for a good crop corresponds pretty closely with that deduced 
from the modern observations of Mr Thomas Chaplin at 
Jerusalem, whence it can be inferred that the climate of 
Palestine has not changed/ 5 

There are a number of old meteorological journals in which 
the wind and weather are given, but no instrumental readings. 
The best known of these are the journal kept by the Rev. 
William Merle at Oxford from 1337 to 1344 (2), and that of 
Tycho Brahe (3) at Uranienborg on the Island of Hveen in 
the Sund from 1582 to 1597. Merle's journal presents a 
picture of the weather which would not differ greatly from that 
given by a similar journal at the present day. The winters 
were certainly not invariably rigorous, for example, 1342 : 
" It is also to be noted that there was spring-like weather for 
the whole time between September and the end of December, 
except on those days to which frost is ascribed, so much so 
that in certain places the leeks burst forth into seed, and in 
certain places the cabbages blossomed.'' Unfortunately, the 
journal is not a day-to-day record so much as a weekly or 
monthly summary of the weather, so that it is not easy to 
extract numerical data like the frequency of rain-days which 
can be compared with similar figures at the present day. 
An attempt to count up the rain-days for the two most complete 
years, 1341 and 1342, omitting only the " extremely light " 
or " very light " rains, gave totals of 152 and 153 respectively, 
compared with a present normal of 168, but the difference is 
of no significance. 

The observations of Tycho Brahe seem to be exceptionally 
favourable for determining a difference of climate between 
the sixteenth and the nineteenth centuries, because the site 
could be accurately identified, and a further series of observa- 
tions was made at the same spot from 1881 to 1898. P. la Cour 
(3) also has made a careful comparison between Tycho Brahe J s 
observations and the mean results at fourteen stations in 
Denmark. The most important difference is that the prevailing 


wind, which is at present from south-west throughout the year, 
was in the sixteenth century from south-east, especially in 
winter. In winter, south-east winds are cold and dry, whereas 
south-west winds are mild and moist, leading to the inference 
that the winters were more severe in the latter half of the 
sixteenth century than they are at present. The number of 
rain-days recorded by Tycho Brahe is about thirty per cent, 
below the present mean in winter, whereas in summer the 
two figures are nearly the same. There is, however, the 
possibility that Tycho Brahe missed some rain-days in winter, 
when he would have been out of doors less than in summer. 
The number of days with snow is greater than at present, 
confirming the view that the winters were colder. H. H. 
Hildebrandsson, however, pointed out (4) that the period 
1582 to 1597 appears to have had severer winters than the 
remaining parts of the sixteenth century ; Tycho Brahe's 
observations happened to coincide with a cold spell and were 
therefore not representative of the century. This conclusion 
from the observations and Hildebrandsson's commentary will 
be fitted into their place in the sum total of evidence concerning 
climatic changes in Europe in the next chapter. 

Observations of wind direction are probably the most 
valuable of all the records of old weather diaries, since they 
can often be compared directly with present-day records. 
An analysis of old wind records in the British Isles, made 
by C. E. P. Brooks and T. M. Hunt (5), presented several 
results of interest (see Chapter XVIIL). If similar studies 
were made for other parts of the world, our knowledge of 
climatic changes would be greatly extended. 

Some weather journals, apparently from Alexandria, 
dating from the early part of the Christian era, described 
by G. Hellmann, are referred to in Chapter XX. These 
journals would be of the very greatest importance in 
demonstrating a change of climate, if it were absolutely certain 
that they were made at Alexandria, and not in Greece. That 
is the chief difficulty in dealing with early meteorological 
observations, whether instrumental or not ; there is generally 
an element of doubt somewhere. 

Still less satisfactory are inferences drawn from early 
descriptions of the climate and physical nature of various 
countries. The Roman writers described Britain as damp 


and cloudy, but so would an Italian of the present day, and 
we are left in doubt as to whether it was any damper or 
cloudier at the beginning of the Christian era than it is to-day. 
The general analysis of the literature of the Mediterranean 
countries initiated by Arago (6), and continued by a large 
number of meteorologists and antiquarians, has shown that 
in these countries during the first century of the Christian era 
the nature of the vegetation and crops, the dates of sowing 
and reaping, and the animal life, all suggest that the climate 
differed little from the present. Arago's remarks about the 
date and the vine have been quoted by every opponent of 
climatic change for the last ninety years the date cannot 
ripen its fruit in a mean annual temperature below 21 C., 
the vine cannot abide a temperature above 22 C. ; since 
both date and vine flourished in ancient Palestine, the mean 
annual temperature must have been 21 C., which is also its 
present value. It seems doubtful, however, whether the 
solution can be quite so simple as that ; differences of exposure 
must come in, and the annual range of temperature from 
summer to winter. Even if Arago's strict limits of temperature 
be accepted, in a country of such varied relief as Palestine, 
the area over which the mean annual temperature at ground - 
level (as opposed to mean sea-level) lies between the limits of 
21 and 22 C. must be quite a small proportion of the whole. 
The effect of a slight change of climate would be nullified by 
moving the plantations to a site with a different exposure or 
at a different level. 

There are two curious features of this mass of anti-variation 
literature started by Arago. The first is that it is almost 
entirely directed against the idea of a progressive change of 
climate, and not against climatic fluctuations. The old 
theory of progressive desiccation has been dead for many years, 
and all this reiteration is merely killing the slain, for to prove 
that the climate of the first century A,D. resembled that of 
the present does not prove that the climate of the seventh 
century A.D. also resembled the present. The reason probably 
is that the progressive theory offers the opportunity for a 
definite negation, while the theory of fluctuations does not. 
The weather of one year differs from that of another year, 
the weather of one decade from that of another decade ; why 
should not the climate of one century differ from that of another 


century ? The question is one, not of fact, but of degree, 
which is much less satisfactory. The second point is that 
practically the whole of the literature is directed against the 
idea that the climate of Europe, Asia, Africa, has become drier. 
No one has attempted to prove that the climate has not 
become wetter, because the fact is so obvious that no proof 
is needed. 

The discussion of the literary records of weather follows 
a different line of argument. There have at all times been 
annalists, who wrote down accounts of the striking events of 
their time. They were not concerned particularly with the 
weather, but if a great flood or drought, frost or storm, occurred, 
they wrote it down. These weather notes have been extracted 
by various commentators, who have often been at great pains 
to verify the dates and eliminate the errors introduced by 
copyists, so that a large amount of fairly reliable material is 
now available. It seems a reasonable argument that if a 
considerable number of droughts were recorded in one century, 
the rainfall of that century was abnormally low ; similarly, 
a large number of floods and storms suggest a heavy rainfall. 
There are, however, several difficulties to be overcome. 
The first is that the completeness of the record changes from 
one century to another. Thus the number of records of 
droughts may be six in the seventh century and ten in the 
thirteenth century, but this does not necessarily mean that the 
latter was the drier. The records of storms and floods may 
number two in the seventh century and twenty in the thirteenth. 
The correct way of stating the evidence would be that of the 
total number of records of raininess in the seventh century, 
25 per cent, indicate a high rainfall ; of those in the thirteenth 
century, 67 per cent., so that the latter century was the wetter. 
This gives us a satisfactory method of dealing with records 
of raininess, but, unfortunately, records of temperature cannot 
be dealt with in the same way, for we have practically only 
records of severe winters or hot summers, the mild winters and 
cool summers being less often recorded. 

The second difficulty concerns the psychology of the annalists. 
Vanderlinden (7), in the preface to his " Meteorological 
Chronicle of Belgium," divides records of this type into three 
stages. In the earliest stage, the authors are concise ; they 
merely state " cold winter,' 5 " dry year," etc., without any 


subjective remarks. Later, the records become longer and 
more fanciful ; often the chronicler breaks into verse. In 
the third stage, there is a certain amount of manipulation of 
facts, under the influence of religious or superstitious ideas, 
and it is not until the end of the eighteenth century that the 
reports again assume a concise and scientific character. 
Finally, we have the difficulty that all these annotations tend 
to be comparative. Suppose that after a long period of dry 
climate there is a change in the direction of greater rainfall, 
which accomplishes itself in a period of fifty years, after which 
the climate continues at its new level. Obviously the period 
of increasing rainfall and the early subsequent years would 
suffer by comparison with the dry period which preceded 
them, while after the rainier climate had prevailed for one 
or two generations it would be accepted as the normal order 
of events and would escape comment. Thus, from the records, 
the period during and immediately following the change 
would actually appear as a rainfall maximum. I think the 
maximum rainfall indicated for Europe in the eleventh century 
is due in this way to the abrupt change from the dry conditions 
of the tenth, while the maximum of the thirteenth century, 
which was probably equally if not more pronounced, hardly 
appears. The oscillation in the ninth and tenth centuries 
also is probably exaggerated from this cause. 

There is some discrepancy between the records of the 
Classical period and those of the Middle Ages, because the 
centre of civilisation moved northward in the interval from 
the Mediterranean to Western and Central Europe, that is, 
from a generally drier to a more humid climate. The 
meteorological events which are considered worthy of record 
by the annalists are those which strike them as most unusual ; 
in the Mediterranean, a dry summer is taken for granted, 
while a wet summer is an event to be recorded ; in North-west 
Europe, where the rainfall is usually sufficient at all seasons, 
a drought is the more noticeable event. From the records, 
one might suppose that the Tiber was more often in flood 
than the Thames ; this is because the floods of the Tiber are 
short-lived capricious affairs due to sudden heavy storms in the 
mountains ; they were considered worthy of record, while 
in the Thames the water rises more gradually but also more 
regularly, and a certain amount of flooding occurs almost 


every winter. For these and similar reasons the early records 
give an appearance of wetness. Hence the climatic curves 
derived from the literary records have to be " calibrated " by 
reference to the records of lake levels, advance of glaciers, etc. 

Difficulties of chronology may also cause compilations 
of historical records to give a false impression of the variations 
of climate. When the exact date of say a drought is uncertain, 
different annalists may assign it to different years. The 
compiler collects all these dates, and quotes a drought for each 
of them. In this way a few months of dry weather may become 
a drought lasting three, four or five years. Even worse, the 
original drought may not have any real existence. Thus 
G. E. Britton (8), in his model meteorological chronology of 
Britain, considers that the famous three-year drought ended 
by St Wilfrid is a later invention to glorify the Saint. For 
these reasons, in this revised edition I have attached much 
less importance to these early literary records than in the 
first edition and I have omitted the diagrams of the frequency 
of different phenomena as more misleading than useful. 

When we go back beyond the written annals, we come to 
the period of tradition. The traditional meteorological event 
which will spring at once to the mind is the Noachian deluge, 
which finds its parallel in the legends of many other nations 
besides the Jews. The Biblical flood was closely similar in 
its details to a Chaldaean legend recorded by Berosus, but 
most peoples of the Near East, including the Greeks and 
Persians, had similar traditions, all of which were probably 
derived, in part at least, from the Chaldaean. Curiously, the 
Chinese have a flood legend which is remarkably like that of 
the Bible. In the Indian version the saving vessel finally 
landed on the loftiest summit of the Himalayas. There is 
also a flood legend among the Aztecs of Mexico. The meaning 
of this widespread tradition is not clear ; if it refers to a single 
event in Mesopotamia it must be very old, probably earlier 
than 4500 B.C. 

Another meteorological legend is that of the twilight of the 
Norse gods, when frost and snow ruled the land for generations. 
This can reasonably be attributed to a great change of climate 
for the worse which occurred about 500 B.C. But since these 
traditional meteorological events can only be interpreted in 
terms of climatic changes with which we are already acquainted 



through other evidence, they are of little or no help in 
elucidating the actual climatic variations ; at most, they can 
serve as a confirmation of other evidence. 

The fluctuations of lakes and rivers form in general the 
most satisfactory evidence for determining changes of climate. 
The levels of the Central European lakes during the period 
of lake-dwellings are the most reliable source of information 
for the long pre-Glassical period in Europe. In Western Asia 
the variations of the Caspian form a useful index of the rainfall 
during the past 1,500 years or so, eked out by scattered data 
from other salt lakes. Lakes without outlet are the most 
satisfactory because they respond readily to changes of rainfall. 
In both Europe and Asia the fluctuations can generally be 
dated by archaeological or historical correlations ; the pro- 
nounced variations of level in the salt lakes of Western North 
America are of less value because they cannot be dated in 
this way. In Europe the variations of rainfall in the basins 
of several lakes are recorded in the thickness of annual layers 
of sediment, and one of these records goes back to 2300 B.C. 
In Africa there is a unique series of actual measurements of 
the levels of the Nile, which are dated to within a year. 
There is also a large body of evidence about the long-period 
variations of level of the Central African lakes, which point 
to large fluctuations of rainfall, but unfortunately these 
can only be dated very roughly by archaeological means. 
Returning to Central Europe, there is a large body of informa- 
tion as to the fluctuations of the Alpine glaciers, including 
some pre-Classical fluctuations which can be traced and 
approximately dated by archaeological evidence. The laws 
which govern the movements of glaciers are not yet fully 
understood, but a succession of snowy winters and cool wet 
summers seem to be most favourable for advance and hot 
dry summers for retreat. 

The evidence afforded by racial migrations as to climatic 
changes depends on the principle that during a period of 
increased rainfall there is a movement of peoples from regions 
which are naturally moist to regions which are naturally dry, 
while during the drier periods the direction of movement is 
reversed, the naturally moist regions being occupied and the 
dry regions more or less abandoned. E. Bruckner (9) showed, 
for example, that emigration from Europe to the United 


States depended on the rainfall. In order that this principle 
may be used to determine the course of climatic variations, 
certain conditions are necessary. First, there must be large 
areas which are on the borderline between aridity and complete 
desert ; these areas must be mostly too dry for extensive 
agriculture, but with sufficient resources to support under 
average conditions a large nomadic population, while a 
succession of dry years renders them almost uninhabitable. 
In close proximity to this arid region there must be a fertile 
well-watered plain, with a long and accurately dated history. 
During dry periods the nomads are driven from their homes 
by lack of water, but they find little difficulty in moving 
from point to point, and the sedentary agriculturists of the 
neighbouring plain generally find them irresistible. It is 
only in Asia that these conditions are fulfilled in perfection, 
the rich plains of the Tigris and Euphrates, the site of a long 
succession of civilised states, having on the one side the semi- 
deserts of Arabia and Syria, on the other side a great dry 
region extending eastwards and north-eastwards as far as 
China. We should expect a period of decreased rainfall to 
initiate a series of great migrations spreading out from the 
dry regions and recorded in the history of the Mesopotamian 
states as the invasions of barbarians. The history of Egypt 
does not give anything like so complete a record, because 
the desert on either side of the Nile valley is too dry, even 
under favourable conditions, to support a nomadic population 
sufficiently large to have made any impression on the might 
of ancient Egypt. The Hyksos conquest of about 1800 B.C. 
is the main exception, but the Hyksos themselves probably 
came out of Asia. The invasions of China from the west 
provide some evidence of climatic fluctuations in the east of 

The principle that tribal movements were mainly due 
to drought is insisted on by H. J. E. Peake in his study of 
the migrations of the Aryans (Wiros, as he prefers to call 
them). Thus he writes (10, p. 157) : " We have seen 
reason for believing that a period of drought, occurring 
some centuries before 3000 B.C., drove some of them towards 
the Baltic. . . . But the great dispersal was about 2200 B.C. 
On this occasion the drought seems to have been more excessive 
or more prolonged, for it is believed that the steppe was left 


for a time almost uninhabited." This evidence, however, 
is circumstantial and needs to be used with care. Consider, 
for example, the four great outbursts from Arabia, the first 
of which occurred during the fourth millennium B.C., the 
second, or Amorite, about 2000 B.C., the third, or Aramaean, 
from 1500 to 1000 B.C. (according to Peake, mainly 1350 to 
1300 B.C.), while the fourth, or Arabian, culminated in the 
Islamitic expansion of the seventh century A.D. No reasonable 
cause other than drought can be assigned to the first three of 
these migrations, but the fourth might be attributed to the 
influence of the Moslem religion, were it not that it began 
some time before the birth of Mohammed. The Arabs of 
the region east of Southern Palestine relate that shortly before 
the days of Mohammed, or somewhere about A.D. 600, a 
terrible and prolonged drought caused untold havoc, and the 
greater part of the tribe migrated to the African coast near 
Tunis (n). We have also to distinguish between migrations, 
in which large numbers of people (men, women, and children) 
moved away from one region and occupied another, and 
conquests, in which the ruler of a strong country imposed his 
government on his weaker neighbours, and established an 
empire. No one would attribute the conquests of Napoleon 
Bonaparte, for example, to unfavourable climatic conditions 
in France in fact, the reverse conclusion could be argued 
more plausibly, namely, that owing to favourable conditions 
the state became powerful enough to dominate its neighbours 
less favourably situated. The latter is in fact the type of 
argument adopted by Ellsworth Huntington in his historical 
studies. In compiling a list of migrations, therefore, we must 
be careful to omit mere military conquests and raids. 

Huntington's contention, as set out in " Civilisation and 
Climate" (12), is that a certain type of climate, now found 
mainly in Britain, France and neighbouring parts of Europe, 
and in the Eastern United States, is favourable to a high 
level of civilisation. This climate is characterised by a 
moderate temperature, and by the passage of frequent baro- 
metric depressions, which give a sufficient rainfall and 
changeable stimulating weather. Now it is well known that 
the great centres of civilisation in the past lay in more southerly 
latitudes than those of to-day, beginning in Egypt, Mesopo- 
tamia, and the Eastern Mediterranean, and then passing to 


Greece and Rome. Huntington attributes these changes in 
the centres of civilisation to climatic changes associated with 
the northward shifting of the belt of cyclonic activity. In 
another volume (13), he gives a detailed comparison of the 
history of Rome during the Classical period with the climatic 
changes deduced from the growth of the Sequoias in 
California (see Chapter XXI.), which he regards as corre- 
sponding very closely with the rainfall of the Mediterranean 
area. If this principle can be maintained, it obviously 
affords a powerful weapon for deducing the existence of 
climatic fluctuations during the historical period in other 
parts of the world, but S. F. Markham (14) has given an 
alternative explanation, namely that the northward migration 
of the centres of civilisation has followed improvements in 
the methods of heating houses, so that civilised activities 
became possible throughout the year in regions with cold 
winters, and had no relation to changes of climate. 

The width of the annual rings of growth of trees in dry 
regions is closely correlated with the rainfall during the 
preceding few years, so that old trees offer valuable evidence 
as to variations of rainfall during their lifetime. The Sequoias 
of Western U.S.A. at present provide the only accurately 
dated evidence of climatic fluctuations in that country previous 
to the settlement by Europeans. The further description of 
the way in which the records are interpreted is postponed to 
Chapter XXI. So far, the method has not been applied 
to any very old trees outside the United States, but there 
seems no reason why it should not be almost equally effective 
in other continents. 

Geological evidence by itself plays only a very small part 
in elucidating climatic changes during the historical period, 
because it is only rarely that geological deposits can be dated 
with sufficient accuracy. There are considerable possibilities 
in the fine seasonally banded clays which have been forming 
in lakes and quiet fiords in the glaciated regions. These clays 
and the associated annual moraines have yielded valuable 
information concerning the rate of retreat of the ice-sheets at 
different stages, and in the post-glacial period they have 
served to date the peat-bogs and raised beaches which have 
supplied abundant evidence of post-glacial climatic changes 
in Scandinavia, as described in " The Evolution of Climate/' 


The great deterioration of climate which marked the beginning 
of the sub-Atlantic period, about 500 B.C., is dated by 
archaeological evidence. 

From all this it will be seen that our knowledge of the 
climatic changes of the historical period has to be drawn 
from a great variety of sources. While this renders the 
task of reconstruction more difficult, it has the advantage of 
offering frequent opportunities for testing the results by 
comparing the conclusions derived from quite different and 
independent sets of data. An example of this occurs in the 
climatic changes of Europe and Asia. The former are 
deduced almost entirely from geological and archaeological 
data and from the literary records, the latter from migrations 
of peoples and from the levels of the Caspian. The rainfall 
curves obtained for these two continents, however, resemble 
each other so closely that the fluctuations portrayed are 
obviously real. 


(1) HELLMANN, G. "The dawn of meteorology." London, Q,. J. R. Meteor. 

Soc. 9 34, 1908, p. 221. 

(2) MERLE, REV. W. " Gonsideraciones temperici pro 7 annis." The earliest 

known journal of the weather . . . 1337-1344. Reproduced and trans- 
lated under the supervision of G. J. Symons. London, 1891. 

(3) LA GOUR, PAUL. " Tyge Brahes meteorologiske dagbok, holdt paa 

Uranienborg for aarene 1582-1597." Appendix til Collectanea Meteoro- 
logica. Kjobenhavn, 1876. 

(4) HILDEBRANDSSON, H. H. " Sur le pr^tendu changement du climat europe"en 

en temps historique." Upsala, 1915. 

(5) BROOKS, G. E. P., and T. M. HUNT. " Variations of wind direction in the 

British Isles since 1341." London, Q,. J> R. Meteor. Soc., 59, 1933, p. 375. 

(6) ARAGO. " (Evres completes." T. 8. Paris, 1858. 

(7) VANDERLJNDEN, E. " Chronique des eVe"nements me'te'orologiques en 

Belgique jusqu'en 1834." Bruxelles, 1924. 

(8) BRITTON, G. E. " A meteorological chronology to A.D. 1450." London, 

Meteor. Off., Geoph. Mem., 8, no. 70, 1937. 

(9) BRUCKNER, E. " Klimaschwankungen und Volkerwanderungen." Wien, 


(10) PEAKE, H. J. E. " The Bronze Age and the Celtic world." London, 1922. 
(n) HUNTINGTON, ELLSWORTH. "The burial of Olympia." London, Geogr. 

J., 36, 1910, p. 657. 

(12) HUNTINGTON, E. " Civilisation and climate." 3rd ed. New Haven, 1924. 

(13) HUNTINGTON, E. " World power and evolution." New Haven, 1919. 

(14) MARKHAM, S. F. " Climate and the energy of nations." London (Oxford 

Univ. Press), 1942. 



WE have no historical records for Europe which go 
3ack much more than half-way to the year 5000 B.C. 
On the other hand, the geological evidence has been 
studied in great detail, and the chronology of the whole 
period since the glaciers commenced their final retreat is 
rapidly being placed on an exact basis. Since the beginning 
of the Christian era, there is a rich European literature which 
provide? a wealth of material. The " official " end of the 
Glacial period in Sweden, according to the Scandinavian 
geologists, is now dated about 6500 B.C. (i), but by this time 
the climate of Central and Western Europe had become 
definitely temperate, and the latest glacial period really ended 
much earlier. The Fenno-Scandian end moraine, dated 
about 8300 B.C., which encloses most of Scandinavia and 
Finland, would give in some ways a more appropriate date. 
After this the rapidly disintegrating remains of the ice-sheet 
can have had little effect on the climate of Europe. The 
" post-glacial " stages are set out in Table 20, with their 
archaeological equivalents in Western Europe, The latter 
is only approximate, since cultures do not appear simultan- 
eously over the whole area. 

The dating of the various archaeological stages is determined 
by the known historical sequence in the Eastern Mediterranean, 
Egypt, and Mesopotamia, and especially the two latter. In 
Mesopotamia, a fixed point is provided by the total solar 
eclipse of 1 5th June 763 B.C., which was recorded in the annals. 
In Egypt, the dates can be approximately fixed by the heliacal 
risings of Sirius and by certain new-moon festivals, but 1580 B.C. 
is the earliest date in Egyptian history which can be regarded 
as certain within a few years. As regards the dating of earlier 
periods, finality is still far from being attained even in Egypt 
or Mesopotamia, and this doubt is added to in dealing with 
events in Europe which can only be dated by associating them 
with some event in the East. 










Climatic stages. 


Dry, becoming 


becoming cooler, 

Cool, wet 


Culture . Vegetation . 


Alder, Oak, Elm 

Late Tardenoisean Peat 




Early Iron 


Oak giving place 
to Pine 



Beech in Central 

Near present 
Table 20. Post-glacial succession, Western Europe. 

The broad early stages of the climatic succession are now 
very well known from innumerable studies of peat-bogs in 
Northern, Central and Western Europe. The successive 
stages can be placed in sequence by the microscopical analysis 
of the pollen grains which are found in peat-bogs and lake 
deposits. Certain pollen grains of some trees and plants 
are almost indestructible, and since they are produced in 
large numbers and scattered by the wind, they are very 
widespread. The technique of their study has been highly 
developed by G. Erdtman in a large number of papers, and 
has enabled a very close comparison to be made of the forest 
succession in different parts of Europe. The absolute dating 
is given by the prehistoric objects which are found in the bogs. 

Generally speaking, the ground laid bare by the retreat 
of the ice was a maze of depressions and ridges. The hollows 
were occupied by lakes and ponds, and the ridges first by an 
arctic flora, which soon gave place to birch, followed by pine. 
By about 7000 B.C. the climate was dry and sufficiently warm 
in summer for the rapid spread of hazel. The rise of tem- 
perature continued, and with some increase of moisture, by 
6000 B.C. all the western half of Europe was occupied by a 
rich forest of oak, alder and elm, the alder being favoured by 


the increasing rainfall. This was the beginning of the 
" Climatic Optimum/ 5 with temperatures up to 5 F. higher 
than the present, permitting forests to grow much higher up 
the mountain sides than is possible now. The heavy rainfall, 
however, favoured the growth of peat, and about 5000 B.C. 
large areas of forest were killed and buried by peat-bogs. 
This phase continued until about 2500 B.C., with gradually 
decreasing temperature, when there occurred a rather puzzling 
change. The peat ceased to grow and in many places the 
surface of the bogs was occupied by forests of pine and yew. 
This was the Sub-boreal, which was formerly supposed to be 
rather warm and very dry. 

It seems certain that at some stage in the Sub-boreal there 
was a prolonged drought. Lakes decreased in area and in 
a few places trees grew on the floor of dried-up lake basins 
below the level of the outlet ; providing definite evidence 
that at these places the rainfall was less than the evaporation, 
and this enables us to estimate the actual rainfall. The 
following table gives estimates for four such lakes in places 
for which I have been able to obtain statistics or estimates of 
the present rainfall and evaporation. In a drier climate 
evaporation was presumably more active than at present and 
I have accordingly increased the evaporation figures by 




Per cent. 











I 5 .6 








Donegal . . 
Sager Lake 

nr. Bremen . 13-3 27 18 67 

Sechof, Lunz 

Austria ... 20-0 56 27 48 

The average of the last four figures in this table indicates a 
rainfall of only about half the present amount in Western and 
Central Europe. This is less than the rainfall of the dry year 
1921 and points to a real and very marked change of climate 
in Europe. 

Botanists, however, have questioned the possibility of a 
dry period of such intensity lasting for over a thousand years, 
G. Erdtman (3) points out that there was a steady development 


of forests throughout the Sub-boreal, and he considers 
that during this period there was a gradual change towards the 
cool moist Sub-Atlantic type, but at the very end there was a 
relatively short period of perhaps 200 years which might 
be described as a dry heat wave, giving place abruptly to much 
cooler and moister conditions. Similarly H. Godwin and 
A. G. Tansley (4) write of southern England : 

" The best opinion of archaeologists and pre-historians 
generally is beginning to question the validity, for these 
islands at least, of the clear-cut conception of a wet Atlantic 
and a dry Sub-boreal period. Evidence of a major climatic 
Atlantic-Sub-boreal transition in Britain, comparable with 
the thoroughly well established Boreal-Atlantic and Sub- 
boreal-Sub-Atlantic changes, is often lacking and there may 
perhaps have been several important alterations of climate 
between say 5500 and 500 B.C. But it is certain that part at 
least of the Bronze Age in England was relatively dry. In 
contrast with Neolithic settlements on chalk summits, chalk 
uplands seem to have been practically uninhabited during the 
Bronze Age, plausibly because of shortage of water. 

In the middle of ist millennium B.C. climate became cool 
and wet ; of the reality of that transition there is no doubt." 

H. Godwin (5) considers that chalk and limestone soils, 
now mostly occupied by natural beechwoods, were too dry to 
carry close woodland during the Sub-boreal and raised bog 
surfaces were much drier than in the Sub-Atlantic. But a 
bog-surface is very sensitive to climatic change, and in some 
British bogs there are several layers indicating a cessation 
and resumption of growth. E. M. Hardy (6) found five dry 
phases in the bogs of Shropshire, which he places in the 
upper part of the full Boreal, in the Atlantic, in the Sub- 
boreal, at the top of the Sub-boreal and about midway in 
the Sub-Atlantic. These would be about 6000 B.C., 5000 B.C., 
1200-800 B.C., 700-500 B.C., and at the beginning of the 
Christian era. 

E. Granlund (7) from an intensive study of Swedish peat 
bogs, also found evidence of five dry layers, which ended 
about 2300, 1 200, 600 B.C. and A.D. 400 and 1200, that about 
600 B.C. being the best developed. This probably represents 
the " Grenzhorizont " of the Sub-boreal which is widespread 
in Europe. 


About 500 B.C. there was a very rapid large-scale climatic 
change. Over large areas the Sub-boreal forests were killed 
by a rapid growth of peat, which was certainly caused by a 
great increase of rainfall and probably also a fall of tem- 
perature. As many of the bogs formed at this time are now 
drying up, the rainfall was probably much greater than at 
present. From this peak it has gradually declined, but with 
a number of oscillations of smaller amplitude. The general 
results of these peat-bog investigations forms the main basis 
of the top three curves of Fig. 31. 

BC. O A.D. 





Get \fral Europe 







Fig. 3 1 . Variations of rainfall in Europe. 

The records of peat-bogs are borne out by the evidence 
of land mollusca as summarised by A. S. Kennard and other 
writers in England. These point to a very wet climate in 
the Atlantic period. The beginning of the Sub-boreal was 
still humid but the rainfall was decreasing and by 1500 B.C. 
the wet period had largely passed. About 1000 B.C. the 
climate resembled the present. 

A detailed study of the changes in level in the Central 
European lakes has been made by H. Gams and R. Nordhagen 
(2). The earliest part of the Neolithic, while warm, was 
decidedly moist, but it appears that the greater part of the lake- 
dwelling period was one of low water and of relatively high 
temperature. It is supposed, in fact, that the lake-dwellings 
were established, not in the lakes themselves, but on peat-bogs 
which are now covered by the waters of the lakes. The 


succession of events has been made out most completely in 
the Feder See basin, but other lakes confirm it. During the 
Neolithic period the lake was smaller than now, indicating 
a dry period which seems to have culminated about 2200- 
2000 B.C. This was followed by a period of somewhat greater 
rainfall, but drier than the present, and somewhere in this 
period was a " high-water catastrophe " a brief regime 
of floods which destroyed many of the lake-dwellings. This 
flood period cannot be far distant in time from the great 
eruption of Bronze Age peoples from the Hungarian plain, 
which probably occurred soon after 1300 B.C., and carried 
the Phrygians into Asia (see p. 320). It is quite likely, in 
fact, that the flood was the stimulus which caused the migration 
to a drier climate. After this moist period the lakes again 
shrank rapidly to a second minimum, about 1000 B.C., at the 
close of the Bronze Age. During this dry period the chief 
settlements were located in moist localities, and agriculture 
was carried on in places now above the forest level and even 
above passes which are now glaciated. Sand dunes on the 
lake shores appear to have been formed mainly in this period. 
Near the end of the Hallstatt period, about 500 B.C., the 
levels of the lakes rose suddenly ; in the Boden See (Lake 
Constance) the rise exceeded 30 feet. Most of the lake 
villages were destroyed, and settlement in the Alps reached 
a minimum, while the occasional remains are concentrated 
in the warmest valleys. The Alpine mountain settlements, 
including even those where mining for metalliferous ores and 
salt was carried on, were abandoned, and Gams and 
Nordhagen remark that this climatic fluctuation had the 
appearance of a catastrophe. From this peak the rainfall 
gradually declined and by the Roman period was very little 
above the present, since in the first century A.D. the Boden See 
was near its present level, and roads were made across the bogs. 
From about A.D. 180 to 350, settlements were again con- 
centrated in the driest localities, indicating a return of moist 
conditions. In the fourth and fifth centuries A.D. many 
German settlements were established on low ground, now 
swampy, and this dry period probably continued until the 
end of the tenth or the middle of the eleventh century, inter- 
rupted in the eighth century by a moist interlude indicated by 
a rise of lake levels. 


Traffic across the Alpine passes, as shown by the transmission 
of culture, became important about 1800 B.C. (8) when the 
Brenner Pass first became traversable, and reached a maximum 
at the end of the Bronze Age and in the Early Hallstatt period, 
or about 1200-900 B.C. The valley settlements of the Late 
Hallstatt period developed independently apparently in 
complete isolation, and traffic across the passes was at a 
minimum. There was a slight revival at the end of the La 
Tene period and in the early Roman Empire (200 B.C. to 
A.D. o), but it was not until between A.D. 700 and 1000 that 
this traffic again developed on a considerable scale. There 
was a re-advance of the glaciers in the western Alps about 
A.D. 1300, followed by a retreat to a minimum extent in the 
fifteenth century. Near the end of the sixteenth century the 
glaciers advanced rapidly and about 1605 they overran settle- 
ments which had been occupied since the beginning of history. 
About the same time the glaciers advanced in the Eastern 
Alps, Iceland, where they almost reached the moraines of the 
Late Glacial stages, and probably in other parts of the world ; 
and the period from 1600 to 1850 has been termed the " Little 
Ice-Age." There were minor maxima of glaciation about 
1820 and 1850 ; since then the glaciers and ice-sheets have 
been in rapid retreat in all parts of the world. A useful 
summary of the advances and retreats of glaciers is given by 
F. E. Matthes (9). 

From Russia we have a remarkable series of measurements 
by W. B. Schostakowitsch (10) of annual layers in the mud 
deposits of Lake Saki, a salt lake on the west coast of the 
Crimea in 45 7' N., 33 33' E., separated from the sea by a 
strip of sandy beach. The total thickness of the deposit is 
several metres, most of the individual layers measuring only 
a few millimetres. The measurements, in tenths of a milli- 
metre, were made partly on photographs and partly on the 
original sections ; the earliest layer is dated 2294 B.C., but 
in some parts of the sections it was difficult to distinguish the 
lines separating the annual layers and there may have been 
some errors in calculating the age of parts of the sections. 

This series is a valuable climatic record, comparable in 
importance with the tree-rings of Western America. The 
variations in the thickness of the layers are almost certainly due 
to variations of rainfall, and in particular of heavy rainstorms 


which would cause rapid run-off. The measurements show 
a number of isolated years with very thick layers five or 
ten times as thick as neighbouring layers, which is consistent 
with this suggestion. A possible connexion with rainfall was 
further examined by comparing overlapping five-year means 
with the variations in level of the Caspian Sea. Although the 
curves show many differences of detail, there can be no doubt 
that their general course, from maximum to minimum and 
back to a second maximum, is similar. Since rainfall is 
presumably an important factor in determining the fluctuations 
of the Caspian, it seems probable that the deposits of Lake 
Saki are also a rough measure of rainfall. The measurements, 
smoothed over 50 years, are shown in the lowest curve of 
Fig. 31. This shows the high maximum just before 2000 B.C., 
but the remaining fluctuations do not resemble those to the 
north and west very closely. We may complete the picture 
for South-east Europe with some data for Lake Ostrovo in 
West Macedonia collected by M. Hasluck (n). From 
historical evidence he concludes that the lake was high in 
Byzantine days and in the eleventh and thirteenth centuries, 
after which it was low from at least 1400 to the early nineteenth 
century. These details fit in excellently with Schostakowitsch's 

When we turn to North-west Europe we find fewer records 
to which exact dates can be assigned, but the evidence fits in 
well with that from Central Europe. About 1200 to 1000 B.C. 
there is evidence of considerable traffic between Scandinavia 
and Ireland, probably indicating a minimum of storminess. 
There is other evidence that the climate of Ireland during the 
Bronze Age was dry and favourable, and that a high civilisation 
developed both there and in Scandinavia. The Iron Age was 
a time of great peat-formation, and peat-beds on the Frisian 
dunes between two layers of blown sand are dated 100 B.C. 
Some peat-bogs in Northern France were formed during the 
Roman period. This was a time of eclipse for the northern 
peoples, who were unable to maintain their high culture. 
In about 120 to 1 14 B.C., a period of especially great storminess 
in the North Sea (Cimbrian flood) caused the wanderings of 
the Cimbri and Teutons, and gave the coast of Jutland and 
North-west Germany its present form (18). About A.D. 500 
there began a second period of high culture, centred on Tara, 


in which Irish learning was greatly esteemed in Europe, 
and a further great outburst of maritime activity. 

The thirteenth century was very stormy in the North Sea, 
and many inroads of the sea were reported in the annals. 
G. E. Britton (12) gives the following frequencies of " marine 
floods " in Britain : 

A.D. looi- 1051- noi- 1151- 1201- 1251- 1301- 1351- 1401- 

1050 1 100 1150 1200 1250 1300 1350 1400 1450 
I I I 4 3 II 2 2 I 

The worst years seem to have been 1176 in Lincolnshire, 
1250, when much of Winchelsea was destroyed, and 1287-8 
when there were three inundations. There was a corres- 
ponding maximum on the coast of Holland. Some of these 
storms did great damage. In the winter of 1218-19 the 
coastal defences of Holland and Frisia were broken through and 
large areas inundated. 

A study of the history of water mills in East Kent, by 
G. M. Meyer (13) points to a period of heavy rainfall in 
South-east England, which began some time before 1087 
(Domesday Book) and ended during the thirteenth century. 
The rainfall about 1303 was much less than in 1217, and is 
probably still less to-day. 

I have already alluded to the remarkable absence of 
mention of ice in the early reports of Norse voyages to Iceland 
and Greenland (Chapter VIII.), and this subject is discussed 
further in Chapter XXII. It seems probable from the 
descriptions that previous to A.D. 1000 the climate of Iceland 
was more genial than to-day ; considerable areas cultivated 
in the tenth century are now covered by ice. Some remarkable 
evidence concerning climatic change in Greenland is discussed 
in Chapter XXI. The Iceland glaciers probably reached 
their maximum extent of the Christian era in the first half of 
the fourteenth century. In the fifteenth and sixteenth 
centuries they retreated, to advance again in the seventeenth 
century, several farms being destroyed about 1640 or 1650. 
Since then there has been a slight retreat. 

The records of ice in Danish waters were collected by 
C. I. H.* Speerschneider (14), who concluded that they show 
no evidence of essentially different conditions in the past. 
The first record of a severe ice-year which he could discover 


was for the year 1048. For later centuries he gives the 
following table : 

Century. Severe Ice. Ice-free. 

1 2th . . 3 

1 3th 3 

1 5th 9 5 

i 6th 19 9 

i yth 26 12 

1 8th 27 33 

1 9th 42 58 

From the very few occasions when the duration of the 
ice could be determined, he forms the following summary : 

Period. Gases. Duration (Days). 

1296-1546 4 49 

1 583- 1 595 5 49 

1619-1674 6 76 

1700-1750 6 51 

In opposition to Speerschneider, I think that both these 
tables point to a period of cold winters in the seventeenth 
century, and the first also suggests that the fourteenth century 
was cold. The detailed records place this minimum of 
temperature in the- first half of this century. 

Before proceeding to a further discussion of the literary 
records, it will be well to summarise the results already gained 
(Table 21). In this table the records are collected in chrono- 
logical order. 

The literary records of Europe provide a great mass of 
information which has been described by various authors. 
The British records were collected by E. J. Lowe (15) and have 
been exhaustively analysed up to A.D. 1450 by C. E. Britton 
(12). Records of British droughts have also been collected 
by G. J. Symons (16). For Belgium an extensive compilation 
has been made by E. Vanderlinden (17). In addition, I 
have made use of a compilation by R. Hennig (18) which 
relates to the whole of Europe, but may be rather uncritical. 
For Britain up to 1450, I have used Britton's work exclusively, 
omitting years with both wet and dry periods and for droughts 
after that date I have relied exclusively on Symons's work. 


His rather uncritical assiduity makes his numbers somewhat 
excessive, and there is a discontinuity after 1450. From 
these sources I have summarised the number of years in each 
half-century with 

(a) Great storms, floods, heavy rain, or wet summers. 

(b) Hot dry summers or droughts. 

These are shown in Table 22. 

Under (a) an attempt was made to eliminate floods or great 
rains associated with isolated thunderstorms in summer. 

It will be observed that all the records increase in frequency 
as they approach the present day ; this is, of course, simply 
because the volume of literature becomes greater. In addition 
to this, certain periods stand out because the records include 
an abnormal percentage of storms and rain, or of drought. 
Thus the period from about 600 to 800 appears to have 
been rather dry and, at the beginning, mild, while 800 to 
950 and 1050 to 1350 were generally rainy. The period 
1701 to 1750 comes out as unusually dry. In order to obtain 
the figures of " raininess " of Europe, the three records were 


5400 Moist and warm. 

5000 Drier and cooler. 

4500 Moist, rather warm. 

4000-3000 Becoming cooler and drier. 

2 200 Very dry especially in Central Europe. 

2000 Rainy period. 

(1275 ?) Short maximum of rainfall (lake villages destroyed) . 

1200-1000 Dry and warm, sea traffic. 

850 Somewhat moister and cooler. 

700- 500 Dry and warm. 

500 Sudden increase of rainfall, much cooler. Begin- 
ning of Sub-Atlantic. 

o Climate similar to present. 

100 Drier and warmer. 

J 8o- 350 Wetter. 

600- 700 Dry and warm. Alpine traffic. 

800-1200 Little ice. Rainfall heavy in Central Europe. 

1200-1300 Great storminess, mild winters, probably rainy. 

1600 Beginning of general advance of glaciers. 

1677-1750 Dry period generally, mild winters. 

1850 Beginning of general recession of glaciers. 

Table 21. Fluctuations of climate in Europe. 








" Raini- 









IOO-5I B.C. 





50 B.G.-O 































. , 

. . 

. . 











. . 







. . 


















6 4 















































































































































































X 5 







. . 









J 7 







Table 22. Numbers of (a) storms and floods, 
(b) droughts, in Europe. 

combined in the following way. First of all the numbers 
under the heading (a) for Britain, Belgium, and Europe in 
each half-century were all added together, irrespective of 
whether or no some of those from different sources referred 
to the same year. It was considered that if a particular year 
appeared as stormy in more than one record, greater weight 


should be attached to it, and this process of simple addition 
answered the purpose. The same process was then followed 
with the records under (b). The number of records under 
(a), i.e., storms, floods, and great rains, entered to each half- 
century was then expressed as a percentage of the total number 
under both (a) and (b). The numbers up to A.D. 600 were 
then smoothed by adding together the records for five successive 
half-centuries and allocating the mean to the middle period 
of the five. 

On the whole the figures for " raininess " obtained from the 
literary records agree with the data from other sources, but 
add little to the latter. The apparently rainy character 
of the years 45 1 to 650 is due entirely to a large number of 
records of floods in the Tiber collected by Hennig for the 
last half of the sixth century, and the true figure for raininess 
should probably be much lower, comparable with the period 
651 to 800. The last part of Table 22 is over-weighted by 
Symons's droughts but there seems little doubt that the period 
round 1700 actually was abnormally dry (see p. 309). 

Actual rainfall records date from the seventeenth century. In 
England a record at Townley, near Burnley in Lancashire, 
commenced in 1677 and continued with intermissions until 
1704. Meanwhile, in 1697, a record was commenced in 
Upminster, Essex, which continued until 1716. From this 
date there was a gap of nine years, after which records began 
at Southwick, near Oundle, in 1726, and at Plymouth in 1727. 
Both these are overlapped by a long record from Lyndon, 
Rutland, commencing in 1737, and from that date there is no 
difficulty in carrying on the story. The records from 1726 
onwards were collated by G. J. Symons who published a 
well-known table of the annual values in British Rainfall for 
1870, but owing to the gap the earlier records could not be 
compared directly with his series, while the Plymouth record 
came to light after Symons had completed his calculations. 
The method employed by Symons was to calculate for each 
station the average for the period over which its records 
extended, and to express the rainfall for each year as a 
percentage of that average. The various series were then 
reduced fo a common basis by means of corrections deduced 
from the years during which pairs of records overlapped. 
This process is sound if the records are truly homogeneous, 


and if the stations are close together and overlap by a sufficient 
number of years, but it becomes hazardous when it is applied 
several times to isolated records and the overlapping periods 
are short. For example, the reduction from the average at 
Southwick to that at Lyndon is based on an overlap of only 
three years, of which 1737 was much wetter at Lyndon than 
at Southwick, while in 1738 and 1739 Southwick was slightly 
wetter than Lyndon. Hence some sort of a check seems to 
be necessary, and this was provided by expressing the early 
figures as percentages of the normal for the period 1881 to 
1915, and comparing these figures with the percentages 
obtained by Symons. The results show very close agreement : 

Percentage given Percentage of 
by Symons. 1881-1915. 

1726-1736. Southwick ... 99 97 

1740-1760. Lyndon .... 78 78 

1727-1752. Plymouth ... 83 84 

This table is a remarkable testimony to the accuracy and 
judgment of the reductions carried out by Symons. It is 
reassuring on another point also. We know nothing as to the 
construction and exposure of these old gauges, but if they had 
recorded less than the correct amounts, the figures in the 
second column of percentages would have been lower than those 
in the first. The figures may be accepted as reasonably 
accurate, and this encourages us to accept also the figures 
for Townley and Upminster. These cannot be reduced to 
the present-day normal by Symons's method, so they have been 
compared directly with estimated normals for the period 
1881 to 1915. 

In this way we obtain the following figures of rainfall 
in percentage of present normal : 

Per cent, of normal. 

1677-1686. Townley 89 

1689-1693. Townley 90 

1697-1703. Townley and Upminster ... 91 

1704-1716. Upminster 93 

1726-1730 100 

The rainfalls for the different decades, revised and brought 
up to date by J. Glasspoole, are given in the second column 
of Table 23, expressed as a percentage of the long period 
average for 1726 to 1940. Figures in brackets indicate an 
incomplete decade. The third column gives values for Paris 
























1 02 



1 08 



































































i 90 i - i 9 i o 










191 1-1920 


1 06 

1 02 











1 08 














Table 23 



by decades, percentages of 



average, Western Europe. 

collected from various sources and expressed as percentages 
of the average for 1771 to 1940. The fourth column gives a 
long series for Zwanenburg in Holland corrected to Hoofddorp. 
The final column of this table gives the means of the three 
series, as far as they go. 

Two long records are algo available for Sweden, namely, 
Uppsala, near Stockholm (1741-1940) and Lund in the extreme 
south (1748-1910), but these differ so greatly in the early 
years that they cannot both be correct. They agree in 
showing a low rainfall before 1 760, but whereas at Uppsala the 
period 1761 to 1820 is recorded as very dry, at Lund the years 
1771 to 1810 are excessively wet. These two series have 
therefore been omitted. 

Table 24 gives figures for Padua and Milan. The record 
for Padua runs from 1725 to 1900, that for Milan begins in 
1764. The Padua series was corrected to Milan to make one 
record. The percentages for Paris, Hoofddorp and Milan 
were calculated by Miss N. Garruthers. 

Years. Per cent. 

1725-1730 (in) 
1731-1740 89 

I74I-I750 102 
1751-1760 112 
I76I-I770 113 
I77I-I780 87 

Years. Per cent. 
1781-1790 91 
1791-1800 99 
1801-1810 104 
1811-1820 103 
1821-1830 99 
1831-1840 105 

Years. Per cent. 
1841-1850 118 
1851-1860 106 
1861-1870 92 
1871-1880 100 
1881-1890 108 
1891-1900 105 

Years. Per cent. 
1901-1910 101 
1911-1920 108 
1921-1930 79 
1931-1936 102 

Table 24. Rainfall means by decades, Milan, Italy. 

The rainfall minimum in the first half of the eighteenth 
century appears to have been widespread. Table 25 gives 
the values for places with a record covering at least ten years of 
the period, as percentages of recent averages. 


Per cent, of i Per cent, of 

Place. Period, normal. 

Uppsala . . 1721-1731 87 


Lyndon . . 1740-1749 77 

South wick . 1726-1736 97 

Upminster . 1701-1716 84 

Plymouth . 1727-1750 94 

Place. Period, normal. 

Berlin . . . 1729-1739 91 

Paris . . . 1701-1750 86 

Bordeaux . 1714-1750 87 

Padua . . 1725-1750 103 

Charleston . 1738-1750 100 

Table 25. Rainfall of first half of eighteenth century. 

The figures for Western and Central Europe are remarkably 
consistent, and point clearly to a persistent dry period during 
these years. 

There is one other interesting feature about Table 23. 
Rainfall maxima occur in 1761-1770, 1821-1830, 1871-1880, 
1921-1930, and possibly 1689-1703, i.e., at intervals of between 
50 and 60 years. The Milan series shows maxima in 1725- 
1730, 1761-1770, 1801-1810, 1841-1850, 1881-1890, and 1911- 
1920, i.e t) an average interval of nearly 40 years. W. H. 
Bradley (see p. 108) found a cycle of about 50 years persisting 
for a period of several million years in the Eocene of the 
West-central United States. The cause is unknown, but if 
this cycle is verified from other records it may be of great 
economic importance. 

Fig. 32 aims at giving a reconstruction of the variations 
of temperature in Western Europe. The uppermost curve is 
based mainly on the estimates of botanists from plant remains, 
especially tree pollen (see p. 296) and is necessarily 
generalised. The lower curve is constructed from the estimates 
of the character of each winter given by C. Easton (19). 
Up to 1 200 records are scanty ; from A.D. 401 to 800 only 
4 to 8 per century and from 80 1 to 1200 only 7 to 17 per 
half century. These refer mainly to severe winters and the 
direct average of Easton's figures is therefore too low. I 
applied a rough correction by plotting the mean of the 
years given against the number of years and drawing a smooth 
curve through them, then reading off the difference between 
this curve and the actual mean. The results show cold 
periods about 900, noo and 1551 to 1700, and warm periods 
about 1175, 1300 and 1725. This curve is repeated in the 
right hand part of the upper curve of Fig. 32. 

Reliable instrumental records of temperature begin about 


the middle of the eighteenth century. Series for Lancashire 
were discussed by G. Manley (20) and compared with those 
for Edinburgh, Oxford, Durham and Stockholm. A. Labrijn 
(21) gives a series for De Bilt, Holland, from 1741 to 1940. 
All these show a series of oscillations of the order of 30 to 40 
years, while the winter temperatures, but not those for other 
seasons, have in addition a steady rise which persisted from 
about 1810 until it was interrupted by the series of cold 
winters in the present decade. This rise of winter tem- 
perature has affected the greater part if not the whole of the 
Northern Hemisphere and is an important climatic fact. 

6 OOP 



BC. O AD. 


400 AD 



/ \ < 



^ -J' 


\~' ^-' 


r\J C7 ^ 

Fig. 32. Variations of temperature in Europe. 
Upper curve, general. Lower curve, severity of winters (Ea^ston). 

A good deal of information about variations of wind 
direction in Britain is available from various sources. An 
interesting paper by Leonard S. Higgins (22) gives some 
inferences as to the prevailing wind directions in South Wales 
since about 400 B.C. The sand dunes have formed where the 
coast faces west or south-west, and have since moved inland at 
intervals. The archaeological and historical evidence, which 
is plentiful, is discussed in detail, with the following results : 

1 . Blown sand was present, and had recently been increasing, 
about 400 to 200 B.C. 

2. After that date the area appears to have become stable, 
and the dunes fixed by vegetation, until the end of the 
thirteenth century. 

3. Soon after 1300, references become frequent to moving 
sand obliterating roads and pastures and burying 


buildings. By 1553 an Act of Parliament " touching 
the sea sands of Glamorgan " had become necessary. 
4. After about 1550, conditions appear to have gradually 
become more stable, and the dunes became increasingly 
fixed by a plant cover. 

The inroads of the sand depend on the conjunction of 
several circumstances, especially a period of abnormally high 
tides and a period of stormy west or south-west winds. The 
results of the investigation, therefore, suggest periods of 
stormy winds from west or south-west before 200 B.C. and again 
from about 1300 to 1550, while during the intervening and 
succeeding periods these winds were less frequent or less 

From 1341 to 1343 there are a number of entries of wind 
direction in the weather diary kept by the Rev. W. Merle. 
These were analysed by G. E. P. Brooks and T. M. Hunt (23) 
and show a dominant wind from west-south-west. 

In the latter half of the sixteenth century, however, the winds 
appear to have been more easterly. G. M. Meyer wrote to 
me : " The following quotation is taken from * A Restitution 
of Decayed Intelligence,' by Richard Verstegan, first published 
at Antwerp in 1605 ? ^ appears on p. 109 of a copy dated 
London, 1634 : "... old shippers of the Netherlands 
affirming, that they have often noted the voyage from 
Holland to Spaine, to be shorter by a day and halfe sayling 
than the voyage from Spaine to Holland." 

The natural inference to be drawn from this record is 
that, during the sixteenth century, easterly winds were more 
prevalent in the English Channel than westerly winds 
contrary to modern experience. The following quotation 
from a book published in 1579 though ambiguous is to 
the same effect : " And the winds in Winter blow about the 
morning, but in the Sommer about the evening, and in 
the Winter out of the East, and also in the Sommer, out 
of the West." 

The observations of Tycho Brahe at Uraniborg in Denmark 
from 1582 to 1597 also point to conditions over North-west 
Europe differing from those prevailing at present. According 
to the discussion of these observations by D. La Cour (24) all 
easterly winds were more frequent than at present, and the 

EUROPE 3 1 3 

prevailing direction was actually south-east instead of 

About the middle of the seventeenth century the winds appear 
to have been rather variable in England. Sir Francis Bacon 
(25, p. 35) wrote : " In Europe these are the chief stayed 
winds, North windes from the Solstice, and they are both 
forerunners and followers of the Dog-starre, West windes from 
the Equinoctiall in Autumne, East windes from the Spring 
Equinoctiall ; as for the Winter Solstice, there is little heed 
to be taken of it, by reason of the varieties." On page 39, 
however, there is an indication that the wind was mainly 
westerly. Probably the winds were more variable than at 
present but with some predominance from the west. 

From 1667 onwards (23) there are a sufficient number of 
old weather diaries to give a fairly complete picture. The 
direction and " constancy " (per cent.) for each half-century 
are as follows : 



Constancy . 


Constancy . 


Constancy . 

The resultant directions are shown in degrees from north 
through east ; 180 is a south wind, 225 a wind from south- 
west and 270 from west. Constancy is the ratio between 
the total movement of the air from the resultant direction, 
assuming that each wind has unit velocity, divided by the 
number of winds and multiplied by 100. A constancy of 33 
means that two-thirds of the winds blew more or less from the 
prevailing direction and the remainder from opposing direc- 
tions. The figures show for London a steady swing from 
nearly W.S.Wi to nearly south, and back again. The 
most remarkable periods at London were from 1740 to 
1747 and from 1794 to 1810, both of which show a 
predominance of easterly winds. The Dublin series also 
shows mainly easterly winds in 1740 to 1748, but observations 
are missing for the second period. The decade 1740 to 1749 





























2 5 8 










was the driest in England since observations began. During 
the period 1794 to 1810 the winters in England were 
exceptionally severe. The years 1901 to 1930 were remarkable 
for the great steadiness of the W.S.W. winds, which is no 
doubt related to the rise of winter temperatures. 

We are now in a better position to examine Huntington's 
contention, set out in the preceding chapter, that the level 
of civilisation in Rome fluctuated in accordance with the 
average rainfall. The vigorous Roman life of the early 
Republic, based on intensive agriculture, was maintained 
during the period from 450 to 250 B.C. Towards 250 B.C. 
the spirit of discipline and rural simplicity began to decay, 
from 225 to 200 B.C. was a period of economic stress, and the 
second century B.C. witnessed a great decline of agriculture. 
During this period malaria became endemic, according to 
E. Huntington (26), because the decreasing rainfall was no 
longer sufficient to maintain flowing water in the rivers 
throughout the long hot summers, so that stagnant pools and 
marshes were formed which provided favourable breeding- 
grounds for mosquitoes. After the land law of Spurious 
Thorius in in B.C., however, agricultural disturbances 
declined and the price of land rose rapidly, but the vine and 
olive replaced grain as the main agricultural product. By 
90 B.C. there was a marked increase in general luxury and 
comfort which reached a high level from 75 B.C. to about 
A.D. 50. From A.D. 80 onwards, however, there was a gradual 
decline, and A.D. 180 to 190 were years of famine and pestilence. 
From A.D. 193 to 210 there was a slight increase in prosperity, 
but then began in full force the long " decline and fall of the 
Roman Empire." 

When we compare these variations in the level of prosperity 
with the estimates of rainfall we obtain the following result : 






IOO B.C.- 

A.D. 50 



A.D. 200 



Very high. 





Very low. 


Very heavy 
until 300, 
then de- 


Light until 
150 B.C., 
then in- 

Heavy at 
first, then 


tLight and 


The agreement is surprisingly good, but the climatic 
changes seem to precede the changes of civilisation by about 
fifty years. This is actually rather more probable a priori 
than a direct concordance. 

If this principle is sound and if the rainfall curve is correct, 
we should expect to find a recrudescence of civilisation in 
the Mediterranean from the middle of the eleventh to the 
middle of the thirteenth centuries. At that time a large part 
of this region was in the hands of the Moslems. It was seen 
in the last chapter that the Moslem outbreak from the seventh 
century onwards was the fourth of a great series of waves of 
emigration from Arabia which are attributed to dry periods. 
The Moslem armies rapidly overran North Africa and Spain, 
and at first their achievements were largely military and 
religious. About the eleventh century, however, they began 
to develop a high civilisation, and Egypt for a time took 
almost its old place as a leader of thought. As an example of 
the high level of Moslem culture in Spain, we have the 
Alhambra at Granada (thirteenth century). Italy also 
reached a high level during this period, when the city states, 
especially Venice and Genoa, became famous for example, 
the cathedral of St Mark in Venice was built in the eleventh 
and twelfth centuries. The agreement seems to afford 
additional support to the rainfall maximum of this period 
in Europe, and it justifies us in using, with due caution, 
variations in the level of civilisation as indications of climatic 
change in other regions also. 

It is with regard to ancient Greece that the discussion of 
Huntington's theory of civilisation and climate has been 
most vigorous. In " The Burial of Olympia " (27), Huntington 
first put forward for discussion the theory that up to about 
400 B.C. Greece had been well watered and forested, with 
perennial streams unsuited to the development of mosquitoes, 
but that after that date the rainfall diminished greatly. The 
streams were reduced in summer to stagnant pools and 
swamps, with the result that malaria became endemic and 
undermined the vitality of the population. The driest period 
began about the seventh century A.D., and resulted in the 
accumulation above the ruins of Olympia of about fifteen feet 
of silt. At present the river Lodon is again cutting a channel 
through this silt. Unfortunately, the hydrographical system 


of this river is so peculiar that it is doubtful whether any 
significance can be attached to this deposit of silt. 

The evidence of the Classical writers is very conflicting, but 
E. G. Mariolopoulos (28) will not admit that there has been 
the slightest change in the climate of Greece since Classical 
times, basing his arguments chiefly on the descriptions of the 
fertility of the country, the nature of the streams and rivers, and 
the dates of sowing and of harvest. He is able to make out 
a strong case against climatic change since at least 350 or 
400 B.C., and perhaps the best verdict is one of " not proven." 
Mariolopoulos gives one interesting quotation from Plato 
which shows that questions of climatic change are not new, 
but agitated the scientific circles of ancient Greece as well as 
those of to-day ; it reads exactly like the report of the recent 
Drought Investigation Committee of South Africa, the 
decrease in fertility being attributed to the washing away of 
the soil. 


(1) ANTEVS, E. " Swedish late-Quaternary geochronologies." New York, 

Geogr. Rev., 15, 1925, p. 280. 

(2) GAMS, H., and R. NORDHAGEN. " Postglaziale Klimaanderungen und 

Erdkrustenbewegimgen in Mitteleuropa." Miinchen, Geogr. Gesellsch. 
Landesk. Forschungen, H. 25, 1923. 

(3) ERDTMAN, G. " Some aspects of the post-glacial history of British forests." 

London, J. EcoL, 17, 1929, p. 42. 

(4) GODWIN, H., and A. G. TANSLEY. " Prehistoric charcoals as evidence of 

former vegetation, soil and climate." London, J. EcoL, 29, 1941, p. 117. 

(5) GODWIN, H. " Pollen analysis and forest history of England and Wales." 

Cambridge, New Phytologist, 39, 1940, p. 370. 

(6) HARDY, E. M. " Studies of the post-glacial history of British vegetation." 

V. " The Shropshire and Flint Maelor mosses." Cambridge, New 
Phytologist, 38, 1939, p. 364. 

(7) GRANLUND, E. " De Svenska hogmossarnasgeologi." Stockholm, Sverig 

geol. Unders., Afh.C., 26, 1932, no. i. 

(8) CHILDE, V. G. " The Danube thoroughfare and the beginnings of civilisa- 

tion in Europe." Antiquity, i, 1927, p. 79. 


Vol. IX., " Hydrology," Chapter 5. 
(10) SCHOSTAKOWITSCH, W. B. " Bodenablagerungen der Seen und periodische 

Schwankungen der Naturerscheinungen." Repr. from Leningrad, Mem. 

Hydr. Inst. See London, Meteor. Mag., 70, 1935, p. 134. 
(i i) HASLUCK, M. " A historical sketch of the fluctuations of Lake Ostrovo in 

West Macedonia." London, Geogr. J., 87, 1936, p. 338. 

(12) BROTON, G. E. "A meteorological chronology to A.D. 1450." London, 

Meteor. Office, Geoph. Mem., 8, No. 70, 1937. 

(13) MEYER, G. M. " Early water-mills in relation to changes in Jjie rainfall 

of East Kent." London, Q,. J. R. Meteor. Soc., 53, 1927, p. 407. 


isforholdene i Danske farvande i aeldre og nyere tid, aarene 690-1860." 
Af C. I. H. SPEERSCHNEIDER. Kjbbenhavn, 1915. 


(15) LOWE, E. J. "Natural phenomena and chronology of the seasons." 

London, 1870. 

(16) [SYMONS, G. J.] "Historic droughts." British Rainfall, 1887, p. 22. 

(17) VANDERLINDEN, E. " Chronique des e*vnements me'teoroiogiques en 

Belgique jusqu'en 1834." Bruxelles, 1924. 

(18) HENNIG, R. " Katalog bemerkenswerter Witterungsereignisse von den 

altesten Zeiten bis zum Jahre 1800." Berlin, Abh. K. Preuss. Meteor. Inst., 
Bd. 2, No. 4, 1904. 

(19) EASTON, C. " Les hivers dans PEurope occidentale." Leyde, 1928. 

(20) MANLEY, G. " Temperature trend in Lancashire, 1753-1945." London, 

d. J. R. Meteor. Soc., 72, 1946, p. I. 

(21) LABRIJN, A. " 200 jaar tempera turwaarnemingen in Nederland." Hemel 

en Dampkr. y Groningen, 40, 1942, p. 41. 

(22) HIGGINS, L. S. "An investigation into the problem of the sand dune 

areas on the South Wales coast." Archaeologia Cambrensis, June, 1933. 

(23) BROOKS, C. E. P., and T. M. HUNT. " Variations of wind direction in the 

British Isles since 1341 ." London, Q,. J. R. Meteor. Soc., 59, 1933, p. 375. 

(24) LA COUR, D. " Tyge Brahes meteorologiske dagbok holdt paa Uraniborg 

for aarene 1582-1597." Appendix til Collectanea Meteorologica. 
Kjobenhavn, 1876. 

(25) BACON, FRANCIS. " The naturall and experimentall history of winds." 


(26) HUNTINGTON, E. " The pulse of progress." New York and London, 1926. 

(27) . " The burial of Olympia." London, Geogr. J., 36, 1910, p. 657. 

(28) MARIOLOPOULOS, E. G. " tude sur le climat de la Grece. Precipitation. 

Stabilit^ du climat depuis les temps historiques." Paris, 1925. 



THE interior of the great continent of Asia has for 
the last twenty centuries or more been occupied 
by nomadic tribes, and we have no great body of 
dated literary records such as that which facilitated our 
study of the climate of Europe during the Christian era. 
On the other hand, south-western Asia includes the sites of 
some of the oldest known civilisations of the globe, and we 
have a rich mine of historical material from which to draw 
conclusions. In Chapter XVIII. we found some evidence 
that even in such a climatically favoured continent as Europe, 
variations in the rainfall from one century to another strongly 
influenced the level of civilisation, and to some extent deter- 
mined the wanderings of peoples. During a period of increased 
rainfall there is a movement from regions which are naturally 
moist to regions which are naturally dry ; this movement 
is noticeable both in the locations of Alpine settlements and 
in the migrations of whole tribes and nations. During the 
drier periods the direction of movement is reversed ; the 
naturally moist regions are occupied and the dry regions are 
more or less abandoned. In the great Eurasian continent 
there is a progressive diminution of rainfall from west to east, 
which extends to the eastern boundary of the region of monsoon 
rainfall in China, and large parts of the interior of Asia are 
on the borderline between aridity and complete desert. 
Hence, it is in Asia that we should expect to find droughts 
recorded most vividly in history, in accordance with the 
principles set out in Chapter XVII. 

Evidence from the earlier periods is given by the semi-arid 
settlement at Anau on the northern margin of Persia. This 
site was occupied from time to time and abandoned during 
the intervening periods, and since there is no evidence of 
conquest, while the periods of abandonment are represented 
by desert formations, it is generally accepted that the inter- 
ruptions were due to drought (i). The dating of the earlier 



settlements is by means of the relative thickness of deposits 
formed above them, the latest is partly historical. The first 
settlement began about 9000 B.C. ; the second, which immedi- 
ately succeeded it, about 6000 B.C. The last part of the first 
settlement and the whole of the second show evidence of 
gradually increasing drought, and the settlement was aband- 
oned soon after 6000 B.C. The site was reoccupied, after 
an interval of desert conditions, about 5200 B.C. This 
third settlement continued until about 2200 B.C. with a short 
interruption, probably due to drought, about 3000 B.C. 
In 2 200 B.C. there began a period of intense drought, and 
about this time not only Anau, but other settlements, such 
as Susa and Tripolje, were abandoned (i). The site was 
not reoccupied until the Iron Age, probably not much earlier 
than Persian times. 

The evidence for climatic changes in Asia since 100 B.C. 
was summarised by Ellsworth Huntington in his last book (2). 
He gives curves of caravan travel in Syria after C. P. Grant 
(3), his own reconstruction of the evidence of lakes and 
ruins in Asia, and a tabulation of the frequency of migrations 
to and from the dry areas since 400 B.C. from A. J. Toynbee 
(4), all of which agree in their main features. They point 
to rainy periods about A.D. 0-200, 400-500, 700-1100, 1250, 
and 1500-1700, and dry periods about 300, 500, noo, and 

The Syrian desert lies across the main land routes between 
Asia and Europe-North Africa, and in its present state offers 
a formidable obstacle even to motor transport. It would 
now be almost impassable by camel caravans, and in dry 
periods these went by a circuitous route. At other times, 
however, they struck boldly across it, which would have been 
impossible without a much heavier rainfall. Huntington 
notes that about 550 B.C. Nabonidus set up his headquarters 
at Terma in north-west Arabia, now a small village, and his 
son sent him couriers and food supplies regularly by camel 
across the desert. 

In order to carry the curve of migrations farther back, 
Table 26 was compiled from three main sources (i, 5, 6), and 
was carried from 5200 B.C. down to the year A.D. 50. It 
will be observed that the migrations are almost always from 
the drier to the wetter regions ; the exceptions are the 


reoccupations of Anau about 5200 and in the first millennium 
B.C., and a wave of migration which began in Central Europe 

about 1275 B.C. and penetrated as far as the east of Asia 
Minor, where it formed the Armenians. 


5200. Anau reoccupied. 

Before 5000. Sumerians and some Semites occupied Mesopotamia. 

4000-3000. First Semitic wave from Arabia. 

3000. Aryans to Baltic. 

2650. Sumer overrun from the north. 

2600. Amorites invade Egypt. 

2450. Kassites, 

2360. Kassites. 

2300-2050. Great dispersal of Aryans (or Wiros). 

2225. Steppe-folk invade Tripolje area. 

2200. Evacuation of Susa and Anau. 

Semites in Mesopotamia. 

2170. Kassites. 

2072. Kassites. 

2045. Kassites. 

2000. Amorites. Tumulus folk. 

1926. Kassites. 

1800. Hyksos conquer Egypt. 

1750. Eruption from Iran. 

1 746. Kassites. 

1700. Aryans into Punjab. 

1500. Aryans. 

1500-1000. Great unrest in Western and Central Asia. 

1350-1300. Aramaean wave from Arabia. 

1275. Migration eastward to Armenia. 

1 1 80. Elamites. 

(1050. Dorians.) 

ca. 600. Anau reoccupied. 

500. Arabs. 

300-250. Break up of Hellenised States in West Asia. 

1 60. Saka. 

A.D. 50. Jafnite wave from Yemen. 

Table 26. List of migrations. 

In order to construct a climatic curve from this table, a 
value was assigned to each century (or other convenient 
period according to the detail of the record) based on 
the " wet-ward " component of migration. The "numbers 
assigned ranged from 5 for the period about 600 (reoccupa- 
tion of Anau, apparently complete cessation of migration 



from the dry regions) to +5 for the great drought about 
2200 B.C. From these numbers the curve shown in Fig. 33 
was constructed up to the beginning of the Christian era. 
The later portion of that curve is derived mainly from the two 
curves given by Huntington (2) and Toynbee's diagram of 
migrations, controlled by the levels of the Caspian (see below). 
The climate of Western and Central Asia during the 
Christian era is best determined from a study of the fluctuations 
of the Caspian and other salt lakes without outlet. The 
level of such lakes is determined by the rainfall and evaporation. 
If the rainfall increases, the level of the lake rises and it over- 
flows its shores until it offers a sufficiently increased surface 
for evaporation to balance the greater rainfall over the basin. 



A.D. 3QO 


WesTar\d Central 







ig' 33- Variations of rainfall in Asia. 

If the increase of rainfall is very great, the lake may rise until 
it finds outlet to the sea. With decreasing rainfall the level 
sinks and the area decreases until the evaporation is again 
only sufficient to balance the rainfall. A decrease in the rate 
of evaporation per unit area would have the same effect 
as an increase of rainfall. Only one of these natural rain- 
gauges, the Caspian Sea, lies sufficiently near to the world of 
antiquity for its variations of level to be brought into our 
chronology, but the variations of the remaining lakes have 
evidently been similar, and we may infer that the fluctuations 
of rainfall indicated by the Caspian represent with fair 
accuracy those of the whole of Central and Western Asia, in 
spite of the complications introduced by variations in the 
course ofthe Oxus River. The fluctuations of the Caspian 
were carefully studied by E. Bruckner (7), and have been 
further discussed by Ellsworth Huntington (8). 


The first definite reference to the Caspian is given by 
Herodotus, about 438 B.C. Huntington interprets Herodotus' 
description as implying that the length of the Caspian from 
north to south was about six times the breadth from west to 
east. At present its length is only between three and four 
times its breadth, but if it became deeper, it would expand 
very little in an east-west direction and very greatly to the 
north. He also considers it probable from the description 
that the Sea of Aral was united with the Caspian. For these 
reasons Huntington believes that when Herodotus wrote, 
the Caspian stood about 150 feet higher than now. Strabo, 
in A.D. 20, gave descriptions from which Khanikof has 
estimated that the Caspian stood at that time 85 feet above 
its present level. On the other hand, L. Berg (9) states that 
these former very high levels are contradicted by the fact that 
deposits containing Cardium edule, still living in the Caspian, 
are found to a height of only 75 feet above the present level. 
These early records, therefore, would not carry much weight, 
unless they were supported by material from other sources. 
This is not the case, for in the discussion on " The Burial of 
Olympia " (Chapter XVIII.) Sir Aurel Stein brought forward 
evidence that about the beginning of the Christian era the 
levels of the salt lakes and marshes in the desert west of Tun- 
huang were about the same as to-day. These form part of a 
defensive line which was completed by a Chinese wall, built 
about 100 B.C. and abandoned early in the first century A.D. 
Wherever this wall abuts on any of the lakes or marshes it 
can clearly be traced down to within a few feet of the actual 
water level in the spring of 1907. 

Nothing further is known as to the level of the Caspian 
until the middle of the fifth century, but J. W. Gregory (10) 
states that in A.D. 333 the Dead Sea stood at its present 
level. Between A.D. 459 and 484 the " Red Wall " was built 
as a barrier against the Huns. At this time the level of the 
Caspian must have been very low, for the wall extends below 
water 18 miles from the shore, and a caravanserai at the old 
port of Aboskun is now under water ; there are other sub- 
merged houses and cities of unknown date in different parts 
of the basin. The level was at least 15 feet below the present. 

Istakhri, an Arab geographer, in A.D. 920, records that 
the old wall at Derbent projected into the sea so far that 

ASIA 323 

six of its towers stood in the water, and from this Bruckner 
concludes that the level was 29 feet above the present. There 
is also evidence that Lake Seistan, in Persia, was high about 
A.D. 900. 

The caravanserai at Baku, which, according to Bruckner, 
was built in the first half of the twelfth century, indicates a 
level 14 feet below the present. In 1306 to 1307 the level was 
37 feet above the present ; this may be partly due to the fact 
that the Oxus entered the Caspian instead of the Sea of Aral 
at that time, but it is significant that almost in the same year 
Dragon Town, on the shores of Lop Nor, was destroyed by 
the rising water. In 1325 the level was still high. 

Early in the fifteenth century the Caspian swallowed up 
a part of the city of Baku (level 1 6 feet above present) . ' The 
level was still high in 1559 and 1562. 

For the seventeenth to nineteenth centuries Bruckner gives 
the following levels : 

1638. 15 feet above pre- 
sent level. 

1715-1720. i foot above. 
1 730- 1814. Relatively high. 

1815. At least 8 feet 

1830. i foot above. 

1843-1846. 2 feet below. 

1847. i foot above. 

1851-1860. i foot below. 

1861-1878. i to 3 feet above. 

above present. 
The changes of level in the Caspian are shown in Fig. 34. 

B.C. A.D. 200 400 00 800 1000 1200 1400 1600 1600 

Fig. 34. Variations in the level of the Caspian. 

The curve of rainfall in Asia since 5200 B.C. agrees well 
with those for Europe (Fig. 31). Both rise to a maximum 
between 5000 and 4000 B.C., falling steadily to a minimum 
about 2200 B.C. The small maximum at 1275 B * G - * s 
dated by evidence of migration from Europe to Asia ; in 
Europe k is supported by other evidence which refers to about 
this period but cannot be dated accurately. The great 
maximum of rainfall about 500 B.C. in Europe appears to 


have come somewhat earlier in Western Asia, and after this 
there is some conflict of detail which may be due to lack or 
misinterpretation of data, or to errors of dating. The 
oscillations in Western Asia appear to have been more pro- 
nounced than in Europe, but the scales of the diagrams are 
only relative, and arid areas are much more sensitive to small 
variations of rainfall than more humid regions. The dry 
period about A.D. 700 on which Huntington lays so much 
stress is also found in Europe, and a general excess of rainfall 
about A.D. 1300 is common to both. Considering that the 
two curves are based on entirely different and independent 
data, the measure of agreement seems to prove that at least 
the major climatic oscillations are real and widespread. 

It should be noted, however, that R. G. F. Schomberg (n) 
does not admit the reality of these variations of rainfall in 
Central Asia, considering that the apparent evidence is 
due to other causes, especially changes in river courses, 
which easily erode the soft sandy soil. In the discussion 
of the last of these papers several speakers questioned this 
verdict. It is, of course, against all climatological experience 
that rainfall could have maintained a dead level for several 
thousand years, but it is not unlikely that Huntington has 
magnified the range of the oscillations. 

There is some evidence of former moister conditions in 
Northern India. E. J. H. Mackay (12), describing excavations 
at Mohenjo-Daru, near the west bank of the Indus about 270 
miles above Karachi in a very dry region, states that about 
2750 B.C. culverts were specially constructed to carry away 
storm water, and between 2750 and 2500 the site was partially 
abandoned because of serious flooding by the Indus. This 
agrees with the paucity of migrations about this time. 
Reference may also be made to some speculations by V. 
Unakar (13) on the interpretation of the RG-VEDA. He 
finds evidence of three types of climate : first a period of 
cool weather with rains fairly uniformly distributed through 
the year and few thunderstorms ; second a stormy period 
when winter depressions gave copious rains ; and finally a 
period of increasing drought. The date is uncertain, but 
Unakar provisionally places the sequence as possibly Covering 
the period 5000 to 2000 B.C. 

A few references given by G. W. Bishop (14) to events 

ASIA 325 

in China throw some light on the climatic changes in that 
country. Thus we have : 

1766 B.C. First dynasty overthrown by a popular revolt 
following seven years of drought. Fig. 33 shows a secondary 
minimum of rainfall in Western Asia about this date, but 
not comparable with the minimum of 2200 B.C. 

1 122 B.C. Second dynasty overthrown by popular revolt 
and invasion from the west. This disturbance clearly 
coincides with a minimum in Western Asia. 

842-771 B.C. Period of turmoil and invasion from the 
west. Great drought, accompanied by disturbances, recorded 
for about 822 B.C. This is in remarkable agreement with 
Erdtman's " dry heat-wave " in the closing centuries of the 
Sub-boreal in Europe. 

Four hundred years of anarchy and confusion began in 
the third century A.D., and this again coincides with a dry 
period in Europe and especially in Central Asia, and a 
shorter period of disintegration in the tenth century also 
falls in a period of drought. 

Co-Ching Chu (15) published an analysis of the Chinese 
archives since A.D. 100 on the same lines as that for Europe 
described in the preceding chapter. His results for all China, 
tabulated by centuries, are as follows : 








































6 4 



1 1 


































Table 27. Floods and droughts in China. 

There is a general tendency for the number of floods to 
increase relatively to the number of droughts in the later 
centuries ; when this is allowed for, the fourth, sixth, and 
seventh centuries and, later, the fifteenth and sixteenth 
centuries stand out as predominantly dry ; the second and 
third, eighth, twelfth, and fourteenth centuries as wet. The 
general agreement with the results of the similar tabulation 


for Europe is very good. The curve of raininess in China, 
based on these data, is shown below the curve for Western 
Asia in Fig. 33. 

Co-Ching Chu also remarks that " In a recent bulletin 
published by the U.S. Department of Labour, Ta Chen has 
found that Chinese migration can be grouped into three 
periods : those of the seventh, fifteenth, and nineteenth 
centuries. During the first period, Chinese migrated to the 
Pescadores and Formosa ; in the second period, to Malaysia ; 
and in the third, about 1860 . . . with destinations in 
Hawaii, North America, and South Africa. Mr Chen found 
that the most significant causes of emigration are pressure of 
population and droughts and famines ; while during the last 
century Chinese emigration was much accelerated by the ease 
of communication and by the demand for labour to open up 
new lands. Such, however, cannot be said of the seventh 
or fifteenth centuries." The dryness of these two periods 
shown in Table 27 is confirmed by these results. 

Co-Ching Chu also gives some records bearing on the 
variations of temperature. The number of severe winters 
per century during the sixth to sixteenth centuries falls to a 
minimum between A.D. 600 and 800, rises to a maximum 
between A.D. noo and 1400, with a well-marked crest in the 
fourteenth century, and subsequently decreases again. The 
variations of frequency, like those of raininess, run closely 
parallel with the variations in Europe. The author also 
examines the dates of the latest spring snowfall in each decade 
at Hangchow during the period 1131-1260, and finds that the 
average date, gth April, is nearly a month later than the date 
of the latest spring snowfall during the period 1905-1914, 
suggesting that the climate was colder and stormier in the 
twelfth and thirteenth centuries than at present, and con- 
firming the evidence of the severe winters. It is also interesting 
to remark that the author finds a parallelism between the 
occurrence of a severe climate and the frequency of sunspots. 

K. A. Wittfogel (16) finds some slight evidence that in 
North China about 1600 to noo B.C. the winter was warmer 
than at present, interest in crops and agricultural rainfall 
starting very early in the year. He thinks that the** summer 
rainfall may have been slightly greater than now. 
* Finally, some reference is necessary to the ruins of Angkor 

ASIA 327 

(17), a lost city, formerly the centre of the powerful and 
highly civilised Khmer Empire, which developed in the 
steaming jungle of French Cambodia, between the Mekong 
River and the frontier of Siam, in about 14 N. Angkor 
flourished between A.D. 600 and 1200, and the empire seems 
to have reached its highest point about A.D. 1000. Climatically, 
the region is very similar to Yucatan, which also had a high 
civilisation, the relics of which are now buried in thick tropical 
jungle (see Chapter XXII. ), though it lies nearer to the 
equator than does Yucatan, and belongs definitely to an ex- 
tension of the equatorial rain-forest belt. The climatic 
conditions of Angkor are very unfavourable for the develop- 
ment of a high civilisation at the present day, and the founding 
of a great city (the population is estimated at a million) in 
such a site suggests a much drier climate about A.D. 600. 
The causes of the break up of the empire and the abandon- 
ment of the city are not known, but it is an interesting 
possibility that about A.D. 1000 or 1050 the climate became 
moister in conformity with the changes in other parts of 
the world, and that the inhabitants fought a losing fight 
against the advancing tide of tropical vegetation for nearly 
two centuries before they finally gave up the struggle and 
migrated to more open country. 


(1) PEAKE, H. J. E. " The Bronze Age and the Celtic world." London, 1922. 

(2) HUNTINGTON, E. " Mainsprings of civilisation." New York and London, 


(3) GRANT, C. P. " The Syrian desert." New York, 1938. 

(4) TOYNBEE, A. J. " A study of history." Vol. III. Oxford Univ. Press, 


(5) " THE CAMBRIDGE ANCIENT HISTORY," vol. i. Cambridge, 1923. 

(6) HADDON, A. C. " The wanderings of peoples." Cambridge, 1919. 

(7) BRUCKNER, E. " KHmaschwankungen seit 1700 . . ." Vienna, 1890. 

(8) HUNTINGTON, E. " The pulse of Asia." Boston and New York, 1907. 

(9) BERG, L. " Das Problem der Klimaanderung in geschichtlicher Zeit." 

Geogr, Abh. hrsg. von. A. Penck in Berlin, 10, H. '2, 1914. 
(10) GREGORY, J. W. "Is the earth drying up?" London, Geogr. J., 43, 

1914, p. 154. 
(n) SCHOMBERG, R. C. F. " The aridity of the Turfan area." London, Geogr. 

J., 72, 1928, p. 357. 
. " The climatic conditions of the Tarim Basin." idem, 75, 1930, 

P- 3 r 3- 

. " Alleged changes in the climate of southern Turkestan." idem, 

80, 1932, p. 132. 

(12) MACKAY, E. J. H. "Further excavations at Mohenjo-Daro." London, 
J. R. Soc. Arts, 82, 1934, p. 206. 


(13) UNAKAR, V. " Meteorology in the RG-VEDA." J. Asiat. Soc., Bombay 

Branch, 9, 1933, P- 53 J io> '934* P- 38- 

(14) BISHOP, C. W. " The geographical factor in the development of Chinese 

civilisation." New York, N.Y., Geogr. Rev., 12, 1922, p. 31. 

(15) CHU, CO-CHING. " Climate pulsations during historic time in China." 

New York, N.Y., Geogr. Rev., 16, 1926, p. 274. 

(16) WITTFOGEL, K. A. " Meteorological records from the divination inscrip- 

tions of Shang." New York, N.Y., Geogr. Rev., 30, 1940, p. no. 

(17) CANDEE, H. CHURCHILL. "Angkor the magnificent." ' (Witherby), 1925. 


THE most important source of information as to the 
variations of rainfall in Africa is provided by the 
levels of the River Nile. As is well known, the Nile 
commences in Lake Victoria, in Central Africa, and flows 
to Lake Albert as the Victoria Nile. From here it continues 
as the Bahr-el-Jebel, becoming known as the White Nile 
after the junction of the Sobat River. At Khartoum it 
receives the Blue Nile, and near Berber the Atbara River, 
both of which originate in the mountains of Abyssinia. From 
the junction of the Blue Nile to the Mediterranean, a distance 
of i, 800 miles, it receives no appreciable accession of water. 
The level of the Nile passes through an extremely regular 
annual variation ; the water is at its lowest in April and May, 
it rises slowly and irregularly in June and the first half of 
July, but rapidly and steadily in the latter half of July and the 
first half of August, remaining high during September and 
commencing to fall rapidly in October. The regular annual 
flood is the source of the fertility of Egypt ; without it the 
whole land would be a barren desert, and hence the levels of 
the flood have been recorded annually, probably for some 
thousands of years. The lowest level reached at the stage of 
low water has been recorded less regularly. Many of these 
records have been lost, but enough remain to form a very 
valuable series, which has been collected and published by 
Prince Omar Toussoun (i). The values of maxima and 
minima at the Roda gauge, Cairo, smoothed by forming 
fifty-year means commencing at successive intervals of ten 
years, are shown in Fig. 35. 

It is necessary to understand exactly the significance of 
both the high and low levels. The White Nile drains a large 
area of equatorial Africa which has a considerable annual 
rainfall ^distributed fairly evenly throughout the year ; more- 
over, it passes through two large lakes, Victoria and Albert, 
which further regulate the flow. Hence the White Nile 




above its junction with the Sobat River discharges an almost 
constant volume of water throughout the year (2). The Blue 
Nile, the Atbara, and the Sobat River, on the other hand, 
originate in Abyssinia, which receives the greater part of its 
rainfall in the summer months. At Addis Ababa the mean 
annual rainfall is 46 inches, and of this amount 35 inches fall 
in June, July, August, and September. Hence it is these 
rivers, and especially the Blue Nile, which supply the waters 
of the annual flood, in which the White Nile plays very little 
part. The Abyssinian rainfall is monsoonal ; Sir Henry 
Lyons (3) showed that it is closely related to the pressure in 

AD 200 


Fig. 35. Levels of Nile. Flood stage and low-level stage. 

the neighbourhood of Cairo, high pressure preceding a low 
Nile flood, and low pressure a high flood. Variations of 
pressure at Cairo are representative of those found over a 
wide area, extending from Beirut to Mauritius, and from 
Cairo to Hong Kong. Pressure variations in this area are 
generally opposite to those in South America and adjoining 
regions, and the fluctuations of the Nile flood therefore 
represent the " see-saw " of pressure between the old and the 
new worlds. The minimum level, on the other hand, depends 
chiefly on the rainfall in the equatorial belt of low pressure, 
which is very closely connected with the intensity of the general 
circulation of the atmosphere. For this reason the level of 
the Nile during the stage of low water is the better guide to 


the general rainfall of equatorial Africa, while the flood levels 
represent the monsoon rainfall of the eastern highlands. 
We should expect the former to show a closer relation to the 
rainfall of Europe than the latter. 

It will be noticed that both maxima and minima show a 
general upward trend. This is due to the deposit of silt, 
which has been steadily raising the level of the Nile bed for 
thousands of years, at the rate of about o i metre per century. 
This is represented by the straight sloping lines in Fig. 35, 
but it is probable that the rate has varied from time to time. 
The steep rise of the minima in the latest years is due to 
artificial control of the water, especially by the Delta Barrage, 
and the data for these years are useless for our purpose. 

The records are made up as follows : there is an almost 
continuous series of records of both high and low levels 
extending from A.D. 641 to 1480. From 1480 to 1830 there 
is a broken record of the flood levels, and the data of low 
levels are very scanty. From 1830 onwards the series are again 
complete, but their value is greatly lessened by the irrigation 
works. There are no connected series of records known 
earlier than A.D. 641, but Prince Omar Toussoun gives 
determinations of the levels which constituted " weak floods," 
" good floods," and " strong floods " in the fifth century 
B.C., and in the first, second, and fourth centuries A.D., as well 
as in the later centuries. From the seventh to the nineteenth 
centuries the mean of these three levels averages o 48 metre 
below the mean annual level for the century, but this difference 
is very variable, with some tendency for a secular increase. 
Hence the means obtained from these weak, good, and strong 
floods can be regarded merely as indications, which point to 
generally good floods in the fifth century B.C., and in the first 
and fourth centuries A.D., and rather poorer floods in the 
second century A.D. 

The maximum about 500 B.C. finds some support in 
Herodotus' "History" (4). Prince Omar Toussoun quotes 
a passage to the effect that unless the flood rose to a level of 
15 or 16 coudees ("cubits," 8-0 or 8-5 metres) it did not 
overflow the fields ; he therefore takes this as representing an 
average ^ood flood. He thinks that these figures refer to the 
" effective flood," that is, the height to which the flood rose 
above the level of low water, for he points out that " the levels 


given by Herodotus, four centuries B.C., are the same (actually 
they are rather higher) as those cited by Ammien Marcellin 
(Ammianus Marcellinus), four centuries A.D. Now it is 
impossible, with the continual elevation of the soil, that 
after an interval of eight centuries these levels should have 
remained the same, if they had for base the same zero* Hence 
it is necessary to consider all the levels mentioned by these 
authors as being effective levels. 55 

I do not agree with this argument, for the irrigation value 
of a flood depends on its gauge level, and not the range 
from low water, and I think the gauge level would be more 
likely to be reported. The figures quoted would agree 
equally well with the assumption that in the fifth century 
B.C. the floods were generally good, while in the fourth 
century A.D. they were generally poor. Even if the figures 
do refer to " effective floods, 55 however, they still point to 
relatively high floods in the fifth century B.C. From the 
seventh to the nineteenth century A.D. the average " effective 
height " of a good flood is 7-4 metres, and the figures quoted by 
Herodotus exceed this by half a metre or more. This piece of 
evidence cannot, however, be regarded as more than an 

Herodotus has another passage (Book II., Chapter 97) 
which may be interpreted in the same sense ; " When the Nile 
overflows, the country is converted into a sea, and nothing 
appears but the cities, which look like islands in the Aegean. 55 
This happens now only when the flood is very high. 

A passage from Pliny, quoted by Prince Omar Toussoun, 
states that the regular flood of the Nile in the first century A.D. 
was 1 6 coud^es (8-5 metres). Whether this refers to the 
actual or " effective 55 flood, it represents a rather good supply 
of flood water, and consequently points to good rains in 

During the period of complete records from the seventh to 
the fifteenth centuries there is a fairly good agreement between 
the flood levels and the low-water stage, although the fluctua- 
tions of the latter are the more violent. Both show a 
minimum about 775, a maximum about 870, a minimum about 
960, a maximum at r 1 10, and a double minimum at 1 220 and 
1300. After 1370 the curves become divergent and generally 
opposed ; it seems as if the figures recorded were mainly the 


highest maxima of the floods and lowest minima of the low- 
level stage. The true rainfall during this period is probably 
to be obtained by a judicious blend of both sets of data. 

There are no other sources of information for Africa which 
can compare in detail with these Nile flood records. The 
climate of the Mediterranean provinces of Africa has been 
exhaustively examined by H. Leiter (5) on the basis of the 
literary references, mainly in the Roman writings. Leiter 
finds no evidence that there has been any appreciable variation 
of rainfall since Roman times, but he thinks there may have 
been a slight rise of temperature during the historic period. 
G. Negri (6) similarly examined the evidence for changes of 
climate in Gyrenaica, and concluded that there had been no 
appreciable change of climate. Both these authors were 
concerned more with the question of progressive desiccation 
than with fluctuations of climate ; they prove fairly conclu- 
sively that the rainfall of this northernmost belt of Africa was 
not much greater during the early centuries of the Christian 
era than it is at present, but they pay comparatively little 
attention to the possibility of prolonged wet or dry periods in 
the intervening centuries. Nevertheless, their careful work 
does seem to show that in this part of the earth's surface the 
climatic fluctuations were probably of smaller amplitude than 
in the regions to the north and to the south, i.e., in Europe 
and at the sources of the Nile. On the other hand, a meteoro- 
logical register kept at Alexandria by Claudius Ptolemaius in 
the first century of the Christian era, and described by G. 
Hellmann (7), strongly suggests a considerable change of the 
summer climate in Northern Egypt since that date. Hellmann 
examines the register in detail, and concludes that there 
is no direct evidence that the observations were not actually 
made in or near Alexandria. The record appears to state 
quite clearly the name of the Observer and the place where 
the observations were made. Fortunately, there is a long 
series of good recent observations in Alexandria with which 
these early observations can be compared. 

This register was re-examined for me by Miss L. D. Sawyer 
(8), who concluded that the disagreement between Ptolemaius' 
register alid recent observations is not so great as is represented 
by Hellmann. The following table is based on her figures, 
converted to percentages : 



Winds from 

N. NE. E. 

SE. S. SW. 

First Century 



2 28 IO 

Sawyer . . 

II O 2 

2 30 8 

W. NW. Variable. 

25 14 8 

Present . . . 35 n 7 4 5 5 5 25 3 

Table 28. Frequency of winds in Alexandria. 

Miss Sawyer points out that northerly winds are most 
frequently referred to in winter, when they are now least 
regular ; in summer there are references to the beginning and 
end of the Etesian winds in Egypt over 40 days apart, during 
which period other winds are mentioned occasionally but no 
reference is made to actual north winds on any of the days in 
that period. This indicates that it is the unusual rather than 
the more commonplace weather conditions that are referred to. 
Hence, while the register indicates that the winds in the first 
century differed appreciably from the present, the difference 
is not so impossibly great as Hellmann makes out. 

Miss Sawyer did not discuss the descriptions of weather 
phenomena, but the same conclusion seems to hold. 

The number of storms is rather high, but such observations 
are of a relative nature. Observations of rainfall occur under 
various designations, most of which are clear. The signifi- 
cance of one term (takas} is not quite clear, but it probably 
means " fine rain." The frequency of rain, as shown in 
Table 29, does not differ greatly from the present frequency 
(though, if " fine rain " is included, the number is rather high), 
but the annual variation, with its entries in summer, is quite 




. Mar. 


May June 

July Aug. Sep. Oct. 

Nov. Dec. 


ist Century 











ist Century, 

" Fine rain " 












1 1 











ist Century 














\ 0-3 


3 o-i 



7 i'5 

Great Heat 

ist Century 



















" Weather 

Changes " 


ist Century 












Table 29. Frequencies of meteorological phenomena in Egypt. 


different from the present. Summer in Alexandria is now 
completely rainless. The distribution of thunder also differs 
from the present, and the annual total is high. The occur- 
rences of " great heat " in the old register also fail to agree 
with the present climate. The constant strong winds from 
north or north-west in summer temper the heat and give this 
season fewer very hot days than the seasons immediately 
before and after the midsummer months. The column for 
" Present " under hot days shows the number of occasions 
during the period 1873 to 1896 on which the hottest day of 
the year occurred in the month in question. 

Very remarkable also are the frequencies of " weather 
changes." At present the period from May to September 
is one of almost uninterrupted fine dry weather. The 
Calendar of Antiochus, dated about A.D. 200, which also 
refers to Egypt, is still more remarkable in this respect, since 
out of fifty-one records of " weather changes," nineteen 
occur in the period May to September. 

These observations point to a climate very different from 
the present summer climate of Alexandria, and resembling 
much more that of Northern Greece. The divergence is, 
in fact, so striking that in spite of the apparent trustworthiness 
of the record, Hellmann considers that there must be some- 
thing wrong, since a climatic change of this degree would imply 
that meteorological conditions differed from the present, 
not alone over Northern Egypt, but over a very wide region, 
if not over the whole earth. The records agree, however, 
with our somewhat scanty records of the Nile flood for that 
period, and with the evidence from Kharga Oasis described 
in the next paragraph, while in Chapter XXII. we shall see 
that the climatic changes did in fact extend over a large part 
of the Northern Hemisphere in a manner agreeing with the 
requirements of meteorology. 

H. J. L. Beadnell (9) has given us an interesting study of 
the probable changes in the water supply of the oasis of 
Kharga, which lies 100 miles west of the Nile valley and 400 
miles south of the Mediterranean, and further details have 
been added by Miss Caton-Thompson and Miss E. W. 
Gardner* ( i o). The water in this oasis is provided by wells, 
the rainfall being extremely small and erratic. 

In Pleistocene times the Kharga depression was supplied 


with water by springs which formed mounds, but it is not now 
accepted that the floor of the depression was ever occupied by 
an extensive lake. In the Neolithic (7000-5000 B.C.) these 
springs ceased to flow, and holes were dug in the tops of the 
mounds to obtain water. The dead or dying springs were 
covered by dune sands. From Early Egyptian to Persian 
times (5000-525 B.C.) the oasis was practically uninhabitable 
owing to lack of water. The Persians replenished the supply 
of water by sinking deep artesian wells. In Kharga itself 
the oldest ruins belong to the time of Darius, about 500 B.C., 
which was a period of great prosperity. 

The oasis continued to be of importance until the 
beginning of the seventh century A.D., but there was a 
temporary decline in the third and early fourth centuries. 
In the seventh century the oasis decayed. No further 
evidence is available until the twelfth century, when the 
oasis was found to be almost depopulated. In A.D. 1225 
Kharga appears to have been more prosperous than it was 
about 1150, and by A.D. 1300 there appears to have been a 
still further improvement. After this we have no information. 

At first sight, this history of the Kharga Oasis seems to be 
very complete evidence of changes of climate in this part of 
Africa, but Beadnell expresses a doubt. The water supply 
is entirely derived from wells in two layers of water-bearing 
sandstone, and the greater part of the supply is artesian, 
rising to the surface under considerable pressure and forming 
flowing wells. These water-bearing sandstones underlie a 
great area in this part of Africa, but it is only in depressions 
that they are at a small enough depth to be tapped. The 
origin of the water is doubtful ; there are three possible sources 
the Nubian reaches of the Nile, the great swamp regions of 
the Sudan, and the rains of Abyssinia or Darfur. Beadnell 
thinks that the Nile is the most probable source. The total 
amount of water in the sandstone is very large, and may 
represent the accumulation of hundreds or even thousands 
of years. During Roman times a large number of wells were 
put down, and these gradually drained the water-bearing 
sandstones, so that the water supply fell off. After the 
Romans left, wells which became choked were not cleaned 
out, and the water supply decreased still further. 

This suggested explanation shows that the variations of 



water supply in Kharga Oasis do not necessarily represent 
synchronous variations of rainfall ; at least the evidence is 
not so convincing as it appears at first sight. Nevertheless, 
the variations in the water supply agree so well with the 
evidence from other parts of Northern Africa that they must 
be due in part to climatic causes, if not in the region of the 
oasis, at least in the region where the water originates. The 
evidence may be set out in parallel columns as shown in 
Table 30, the last column being in part an anticipation of 
later paragraphs. 


Kharga Oasis, 


Other localities. 

500 B.C. 

A.D. 200 

A.D. 4OO 

A.D. 7OO 

A.D. II5O 

A.D. 1225 

Extensive well boring. 

Temporary decline of 


Floods high. 

Floods good. | Alexandria rainier in summer. 

Floods poorer. 

Floods good. 

Great decline. 

Almost depopulated. 

Mandingan Empire, A.D. 320- 

680 (see below). 

Minimum level, Mandingan Empire broke up. 
A.D. 700-1000. j 

Rise of low- 1 
level stage ca. j 
A.D. 1 100. 

More prosperous. j Level low. 

A.D. 1300 ! Still further improve- j Level low. 
! ment. ! 

Traffic in now waterless 
Eastern Desert. 

Prosperous Sudanese States. 

Table 30. Variations of climate in Northern Africa. 

The chief disagreement is between the Kharga Oasis and 
the levels of the Nile from A.D. noo to 1300. If the evidence 
derived from the Kharga Oasis for this period stood alone, 
one would have no hesitation in rejecting it. It has, however, 
a certain amount of support from other regions. The figures 
of raininess in Europe, after a temporary maximum about 
A.D. 1075 fell to a minimum in 1175, rising again to a maxi- 
mum about 1325. Similarly, in Asia the Caspian (Fig. 34) 
appears to have reached a low level about 1150, rising again 
to a maximum in 1300. Mr G. W. Murray informs me that 

i 22 


there is evidence of considerable pilgrim traffic across the 
Red Sea to Jeddah in the thirteenth century from a port 
afterwards abandoned for lack of drinking water. At about 
the same time we have the prosperous Mossi States of the 
Sudan (see below). All this seems to confirm the Kharga 
evidence. It is possible that at this time the fluctuations in 
the equatorial regions were following a different regime from 
those in the north temperate rainfall belt. The levels of the 
Nile in Fig. 35 show that while the fluctuation^ of the Nile 
flood from A.D. 800 to 1300 were generally similar to those of 
the low-water stage, they were on a very much smaller scale. 
It has been pointed out that the low-level stage represents 
mainly the fluctuations of the equatorial rainfall, while the 
flood level represents the rainfall of Abyssinia, and this 
difference of scale may possibly imply that the equatorial 
fluctuations at that time died out northwards, and that the 
northern part of the continent came under the influence of 
the north temperate fluctuations. It is evident, however, 
that further data will be required before the climatic fluctua- 
tions in this part of the world can be set out in reliable detail. 

The Sahara itself is at present occupied only by a few 
wandering tribes. There is, however, some historical evidence 
that during at least one, and possibly two historical periods 
the level of civilisation in the desert was appreciably higher 
than at present (u). The first period is that of the 
Mandingan Empire. The Mandinke are Sudanese negroes, 
who, according to native tradition, maintained a Saharan 
Empire from about A.D. 320 to 680. After this date the empire 
broke up, and it was not until the thirteenth century that the 
second and more authentic cultural period began. Early 
in the fourteenth century " the greatest Sudanese State of 
which there was any authentic record " was centred at Mali, 
in French Guinea ; about the same time the Mossi formed a 
powerful state in the great bend of the Niger, and the Housa 
State was developed at Kano in Nigeria. The Moslem 
Empire of the Central Sudan spread over a large part of 
the Sahara in the thirteenth century, and increased rapidly 
in importance until it reached its highest pitch about 

The shore-lines of the enclosed Central African Lake 
basins of Nakuru-Elementeita and Naivasha, according to 


E. Nilsson (12) point to a number of post-glacial fluctuations. 
At the end of the last main pluviation (Lake III.) the lakes 
dried completely. The first post-pluvial lake (IV.) reached 
almost to the depth of Lake III. but persisted for a shorter 
time. The lakes again dried completely ; after they re- 
formed (Lake V.) there were minor oscillations, superposed 
on a gradual fall, but no complete drying out. Lake V. is 
associated with Leakey's " Gumban A and B " cultures 
(Nakuran) (13). A Gumban A site has been found in the 
deposits of Lake V. The Gumban B burial was near the 
lake and contains fish bones, and Leakey considers that it 
was of about the same date. It yielded a bead which cannot 
be earlier than 3000 B.C. and was probably later. In 1931 I 
thought the Nakuran lake must represent the Sub-Atlantic, 



&ot AFrica 

Fig. 36. Variations of rainfall in Africa. 
(In the curve for East Africa the dates of maxima are conjectural.) 

which was then placed about 850 B.C. (but is now dated about 
500 B.C.) and this correlation was accepted by the archae- 
ologists. Working backwards, the dry stage between lakes 
IV. and V. then represents the " Climatic Optimum." 
Nilsson, however, correlates Lake IV. with the Neolithic 
of Egypt, about 6000 to 5000 B.C., i.e., roughly the Atlantic 
stage of Europe. The Gumban A and B are not certainly 
contemporaneous and the latter may possibly represent a 
high-level stage of the lake subsequent to the highest level. 
If we date the Nakuran as late as 500 B.C. there is a long 
interval after the Late-glacial which is not accounted for. 
In the lower curve of Fig. 36 I have adopted Nilsson's cor- 
relation but extended the rainy period to about 1000 B.C., 
and indicated the main arid period as ending about 5500 B.C. 
This leaves the Sub-boreal and Sub-Atlantic of Europe 


represented only by minor oscillations of the lake levels. 
This is quite possible, since the Sub-boreal is not now regarded 
as very dry and the Sub-Atlantic was not associated with a 
marked advance of the glaciers and ice-sheets. The dating 
of this curve is to be regarded as conjectural. 

On this interpretation. Lake VI., which appears to have 
been quite important, represents the Sub-Atlantic, and Lake 
VII. may date from the thirteenth and fourteenth centuries. 
There is some evidence (14) that Lake Tanganyika was at a 
much lower level less than 1,300 years ago, when it consisted of 
two lakes separated by an isthmus. The natives have a legend 
of the submergence of this isthmus and the joining of the lakes. 

G. W. Hobley (15) has collected a certain amount of 
evidence as to climatic changes in East Africa. Much of 
this is purely geological, and no dated historical evidence 
is given. Most interesting is the reference to Jubaland, 
where there are large numbers of artificial mounds, some 
30 feet high, believed to be funeral mounds of an extinct 
race. fc In addition the large number of well-excavated 
wells, often over 40 feet deep, and the traces of artificial 
dams, all go to prove that this area, which is now practically 
a desert, once carried a large and organised population." 

The coast of East Africa from Mombasa northward is 
studded with ruined towns of the Mahommedan period, 
but their climatic significance is not obvious. 

Farther south we have the ruined cities of Mashonaland, 
of which the best known is Zimbabwe. These are now 
attributed to native construction in the fourteenth century, 
but they may be a reflection of outside influences rather than 
a product of high local culture due to favourable climatic 


(1) TOUSSOUN, Prince OMAR. " Mmoire sur Thistoire du Nil." Le Cairc, 

Mem. Inst. Egypt, vol. ix. 

(2) LYONS, H. G. " The physiography of the River Nile and its basin." Cairo, 


(3) LYONS, H. G. " On the relation between variations of atmospheric pressure 

in North-east Africa and the Nile flood.** London, Proc. R. Soc., A. 76, 
1905, p. 66. 

(4) HFRODOTUS, The History of. Transl. by GEORGE RAWLINSON, 2 Vols., 


(5) LEFTER, H. " Die Frage der Klimaanderung wfthrend geschichtlicher Zeit 

in Nordafrika." Wien, Abh. K. K. Geogr. Gesellsch., 8, 1909, p. i. 


(6) NEGRI, C. " Sul clima della Libia attraverso i tempi storici." Roma, 

Mem. Ace. Nuovi Lincei, ser. 2, vol. i. 

(7) HELLMANN, G. " Uber die Agyptischen Witterungsangaben im Kalendar 

von Claudius Ptolemaeus." Berlin, Sitzungsber. preuss. Akad. Wiss., 13, 
1916, p. 332. 

(8) SAWYER, L. D. " Note on Egyptian winds in Ptolemy's * Prognostics ' 

and Hellmann's criticism of them." London, Q,. J- R* Meteor. Soc., 
57, 1931, p. 26. 

(9) BEADNELL, H. J. LLEWELLYN. " An Egyptian oasis. An account of the 

oasis of Kharga in the Libyan desert, with special reference to its history, 
physical geography, and water supply. " London, 1909. 

(10) CATON-THOMPSON, G., and E. W. GARDNER. " The prehistoric geography 
of Kharga Oasis." London, Geogr. J., 80, 1932, p. 371. 

(i i) KEANE, A. H. " Man, past and present." Rev. ed. Cambridge, 1920. 

(12) NILSSON, E. " Quaternary glaciations and pluvial lakes in British East 

Africa." Geogr. Ann., Stockholm, 13, 1931, p. 249. 

(13) LEAKEY, L. S. B. " The stone age cultures of Kenya Colony." Cambridge, 

Univ. Press, 1931. 

(14) THEEUWS, R. " Le lac Tanganyika." Mouvement Gt'ogr., 33, 1920, col. 

625 ; 34, 1921, col. 49. 

(15) HOBLEY, C. W. "The alleged desiccation of East Africa." London, 

Geogr. J., 44, 1914, p. 467. 



OUR chief source of information about the climatic 
changes in North America is the rate of growth of 
the " Big Trees " or Sequoias of California. Some 
of these trees are of astounding age, and carry our records 
back long before the Christian era. There have been 
difficulties, of course, and the close comparison and averaging 
of the records of a large number of trees have been required 
to give a reliable record of the rate of growth. Douglass has 
used these records very effectively in investigating rainfall 
periodicities in California, but for the discussion of long-period 
climatic fluctuations certain corrections are necessary, and 
these are difficult to determine. Huntington (i) believes that 
the course of climatic variation is the same in California as in 
Central Asia, and he accordingly employs the levels of the 
Caspian for the final calibration of his curve of tree growth. 
This method, however, is fraught with danger ; if one wishes 
to compare variations of climate in America with those in 
Asia, the American curve should, if possible, be derived 
entirely from local evidence. The material for such a 
discussion on American evidence only is provided by a 
valuable Carnegie Institution publication entitled " Quaternary 
Climates " (2), a collection of papers by J. Claude Jones, 
Ernst Antevs, and Ellsworth Huntington. In this volume 
Antevs gives the results of a reinvestigation of the tree-growth 
data, based on all Huntington's measurements (451 trees) 
corrected for age by intrinsic evidence only. 

Huntington's curve, even after correction for age, longevity 
and " flaring " (see p. 345) shows a much greater variability 
in the earlier years than in the later, which is most probably 
due to the much smaller number of trees. I accordingly 
applied a scale correction which decreased from the Beginning 
to the end of the curve and was designed to reduce the 
variability about the mean to approximately the same value 



This corrected curve is the 


uppermost in 


Fig. 37- 

The middle curve is that given by Antevs (2) for trees in 

dry situations, slightly smoothed to bring out the more lasting 

Antevs first divided Huntington's material into two groups, 
trees growing in dry situations and trees growing in wet 
situations. The width of the rings in each decade was plotted 
separately for each tree or, in some instances, for small groups 


500 B.C. 

A.O 5OO 



Fig. 37. Variations of rainfall in U.S.A. 

of trees of the same age. A smooth " middle line " was then 
drawn through the graph, while the maxima and minima 
were connected by drawing smooth " tangents " on either 
side of this middle line. The fluctuations were then reduced 
to the same basis by dividing the distance of any point on the 
graph from the middle line by the distance between the tangents 
at that point. Since it proved difficult to draw tangents for 
the earlier portions of the curve, when the trees were young 
and growing rapidly, only the middle line was drawn for these 
parts of the curves, the distances from this line being used 
without correction. For this part of the curves, before 200 
B.C., the fluctuations therefore appear to be greater in 
magnitude than for the later part. 

The results for the different trees or groups of trees were 
then added together, a correction being applied to allow for 
the incoming of successive groups. 


The curves based on trees growing in dry and moist 
situations show good agreement after A.D. 800, when they 
are based on a large amount of material. The chief difference 
is the retardation of the maxima on the " dry " curve. 
Previous to A.D. 800 the agreement is not so good, presumably 
owing to the smaller amount of material. The differences 
are largely in the minor fluctuations, and when the curves 
are smoothed a better agreement is obtained. 

These curves cannot be regarded as measures of the rainfall 
only ; they must include other factors of tree growth such as 
temperature and sunshine. These factors themselves, however, 
are presumably related in some way to rainfall ; for example, 
when sunshine is abundant, temperature is high and rainfall 
small. Huntington (2, p. 162) gives the results of correlating 
the rate of growth of 112 Sequoias with the rainfall of 
Sacramento, 1863 to 1910, the rainfall season being taken as 
July to June. The correlation between the annual tree 
growth and the rainfall of the immediately preceding season 
is small (+0-13), but when the two preceding seasons are 
added together the coefficient becomes +'22. The re- 
lationship becomes closer the longer the period preceding the 
tree growth over which the rainfall is summed, and the 
correlation between the annual growth and the total rainfall 
of the ten preceding years is +0-58. Thus the curves of tree 
growth evidently reproduce with a fair degree of accuracy 
the variations of rainfall from one decade to another. In 
plotting the corrected curves of tree growth in Fig. 37, the 
values have been assigned to a date five years before the year 
in which the growth ring was formed, in order to allow for 
this lag. 

The method by which Antevs corrected his curves obviously 
tends to eliminate all fluctuations of long period, and the 
curves cannot be expected to show the major fluctuations of 
rainfall similar to that between the seventh and eleventh 
centuries in Europe. A better measure of the long-period 
variations is probably given by the data corrected by 
Huntington's method (i), but omitting the " Caspian correction 
factor." Two corrections are applied, for age and for 
longevity. Trees grow at different rates according to their age, 
young trees usually growing rapidly and old trees slowly. 
Trees which are destined to have a long life usually grow more 


slowly at first than their neighbours which are likely to die 
much sooner. 

The " corrective factor for age " was readily obtained by 
averaging the rates of growth of a number of trees in the 
corresponding years of their lives. In this way the climatic 
and other peculiarities of the individual years (which come 
at different times in the lives of different trees) are eliminated ; 
it is found, for example, that the average growth of trees i year 
old is o- 10 inch ; 10 years old, o- 15 inch ; 40 years old, 0-20 
inch ; 100 years old, o- 10 inch ; 200 years old, 0-05 inch ; 
and so on. The growth curves of the individual trees are then 
corrected for age by dividing the first year's growth by o- 1, the 
tenth year's growth by 0-15, and so on. 

The " correction for longevity " was obtained in a similar 
manner, the average rate of growth of trees which when felled 
had lived for 100 years, 200 years, 300 years, and so on, being 
plotted against the number of years of growth and a smooth 
curve drawn through the points. 

There is another source of error, namely, that due to 
c< flaring " and " buttressing " at the base of the trees. 
" Flaring " means the spreading out of the base of the tree 
so that instead of descending nearly vertically, the trunk 
meets the ground at an angle, like a cone. The correct 
width of the growth-ring would be that measured at right 
angles to the surface of the trunk ; since the measurements 
were actually made on a horizontal surface, the widths found 
are too great. " Buttressing " means that the cross section 
of the trunk of an old tree is not circular, but develops pro- 
tuberances and furrows which add to its strength. The 
measurements were more easily made across the buttresses 
than across the furrows, and this again tends to increase the 
apparent growth in the later portions of the curve. 

The later portions of Huntington's curve can be checked 
by two incidental facts mentioned by him. He states (i) 
that in moist places there are plenty of young trees of all ages, 
but on dry mountain slopes, while there are plenty of mature 
trees 500 or more years old, there are no young trees except 
an occasional seedling or tree of three or four years' growth. 
This suggests that the climate has been drier than that of 
to-day since about A.D. 1400, and that previous to that date 
it was considerably moister. The other fact is that farther 


north, on the shores of Mono Lake, the rings of growth of a 
tree killed by the rising salt water show that it had been 
growing since 1775. This proves that the level of the lake has 
not been as high as it is at present since at least 1775- Both 
these facts fit in with the later portion of curve and confirm 
the correction for " flaring." 

J. Claude Jones (2) has discussed the variations of level of 
the remnants of Lake Lahontan in the Great Basin of Western 
U.S.A. This was one of the great lakes formed in Western 
America during the Quaternary glaciation ; at its maximum 
it covered a continuous area of about 8,500 square miles. 
As the level fell it split up into a number of separate lakes, 
and at present the old basin of Lake Lahontan is occupied by 
large desert " playas " with several small lakes in depressions 
Humboldt, North and South Carson, Pyramid, Winnemuc^a, 
Walker, and Honey Lakes. The discussion centres mainly 
on Pyramid and Winnemucca Lakes. Jones gives a variety 
of observations of the deposits of calcareous tufa, etc., but his 
interpretation of them is obviously erroneous, since they lead 
him to the conclusion that the mastodon and the camel lived 
on in North America into historic times. He has confused 
recent phenomena with those belonging to the Quaternary 
pluvial period, and it is necessary to go over his work and sort 
out the data from the two periods. 

Calcareous deposits or tufas are widely distributed over 
a large part of the old basin of Lake Lahontan. The tufa is 
in three forms, lithoid (stony), dendritic, and crystalline. 
The lithoid and dendritic forms are found at all levels in the 
basin, and appear to have been formed by the activities of 
algae ; the crystalline form occurs as a mineral (thinolite), 
which seems to have been altered from crystals of aragonite 
deposited from a saturated solution of calcium carbonate ; 
it occurs only in the lowest levels of the basin. Evidently, 
it was not until Lake Lahontan had dwindled almost to its 
present small remnants that the water became sufficiently 
saturated for calcium carbonate to be deposited directly 
without the agency of plant life. There seems no doubt, as 
E. Antevs points out in the second memoir of the collection, 
that the mass of the calcareous tufa, including the ithinolite, 
was formed during the shrinkage of Lake Lahontan after the 
Quaternary expansion ; when the thinolite was deposited, the 


lake must have been intensely salt, but at present Pyramid and 
Winnemucca Lakes are only slightly salt, the salinity in 1882 
being 0-35 per cent, in Pyramid and 0-36 per cent, in 

If we know the total amount of salt in a lake and the average 
amount carried in by the rivers in the course of a year, we can 
calculate the period in years since the lake was fresh. As 
Pyramid and Winnemucca Lakes are separated only by a low 
divide, and both receive branches of the same river, the 
Truckee, a very small expansion would suffice to unite them 
in a single lake, so that for the purposes of the discussion they 
can be treated as one. Jones gives four separate determina- 
tions of the age of the system. The first two depend on the 
method described above, of dividing the total amount of 
saline matter present in the lakes by the annual contribution 
of the river, but calculating the age from the chlorine and 
sodium separately (2, pp. 28-29). Jones writes : 

Using a detailed map of Pyramid and Winnemucca Lakes, it is 
possible to obtain the volumes of the lakes by determining the areas 
of the sub-lacustrine contours by means of a planimeter and cal- 
culating the volumes of the respective sections. Such a determina- 
tion indicated the amount of water present in Pyramid Lake as 
7-787 cubic miles and 1-142 cubic miles in Winnemucca Lake. 
The Truckee River has an average flow based on measurements at 
Vista, a station in the Truckee Canyon below all the larger tribu- 
taries, made during the years 1899 to 191 1 inclusive, of 0-274 cubic 
mile per year. It would take Truckee River 28-42 years to supply 
the water at present in Pyramid Lake, and 4-17 years additional to 
fill Winnemucca Lake. But Pyramid Lake contains 1,455 P ar ts 
per million chlorine, Winnemucca Lake 2,184 parts per million, 
and the Truckee only 13 parts per million. ... It would there- 
fore take the Truckee 3,180 years to supply the chlorine in Pyramid 
Lake and 701 years additional to furnish that of Winnemucca Lake, 
or 3,881 years for both. A similar calculation, using sodium instead 
of chlorine, gave 2,447 years necessary, and the other substances gave 
still lower results. Of these calculations the first is probably more 
nearly the truth, as chlorine is the least likely to be removed from 
solution. No great degree of accuracy can J3e claimed, for many 
factors may have influenced the result. While the Truckee River 
is the only stream of considerable volume that flows into the lakes, 
yet a considerable amount of water is supplied by the intermittent 
streams and springs about the borders. . . . The amount of salts 
carried probably varied somewhat with the increase in flow of the 
river . . . although . . . the present data indicate no very great 
change. Of the factors mentioned, only one, the last, would tend 


to make the period greater, while the others would cause the actual 
period to be less than the calculated duration. . . . 

The question may be approached from an entirely different angle. 
As it happened in 1913, the level of Pyramid Lake was but 5 inches 
below the level at the time of Russell's visit. A sample was col- 
lected near the locality where he obtained his southern sample, and 
the chlorine determined. The gain during the 31 years that had 
elapsed between the two visits was 23 parts per million. As the 
lake had essentially the same volume in both instances and the 
samples were taken at the same locality, the variable factors are 
eliminated as far as possible. Dividing the total chlorine found in 
1913 by the gain and multiplying the result by 31, the years that had 
elapsed, gave i ,956 years as the time necessary for the chlorine to 
accumulate, providing the present conditions had not been 
materially changed. 

This method is open to the criticism that it depends on but two 
analyses, and while it is of value as corroborative evidence, yet it 
cannot be considered as conclusive. 

Still another method, one used by Russell, may be employed. 
Knowing the total amount of chlorine in the two lakes and the rate 
of evaporation, the length of time necessary to evaporate enough 
water to supply the chlorine may be determined. Using the recent 
analyses, Pyramid Lake contains 1,440 parts per million of chlorine. 
Assuming that the water carried into the lake was as fresh as in the 
Truckee River, the water before evaporation contained 13 parts per 
million of chlorine. This would make it necessary to evaporate 
1 1 1 cubic miles of river water to concentrate i cubic mile of 
Pyramid Lake water, or, since the lake contains 7 787 cubic miles, 
864-36 cubic miles have been evaporated since the beginning of 
Pyramid Lake. Similarly, 180-43 cu bic miles additional would be 
required to furnish the chlorine in Winnemucca Lake. 

The loss of water from the surface of a lake by evaporation 
depends on two factors, the rate of evaporation and the area 
of the lake. With regard to the rate of evaporation, no actual 
measurements are available for Pyramid or Winnemucca, 
but Jones gives determinations obtained in three different 
ways : the evaporation from open pans on land at Fallen 
averages 65-14 inches a year, and Bigelow, from observations 
near Reno, concluded that the evaporation over open water is 
about five-eighths of that from a pan on land, giving the 
evaporation from the lakes as 40-7 inches a year. Salton 
Sea, formed by a break of the Colorado River, but now 
receiving no appreciable supplies of water, has been falling 
at the average rate of 55*6 inches a year. Finally, the 
average inflow of the Truckee River into Pyramid and 


Winnemucca, divided by the present areas of these lakes, 
gives an annual evaporation of 52 inches. From these three 
determinations, Jones estimates the average evaporation as 
about 50 inches a year. If the lake had always maintained 
its present area of about 370 square miles, the concentration 
of the chlorine would require 4,300 years. Jones thinks, 
however, that the average level of the lake has been higher 
than the present level, and since there is a shelf cut in the rock 
1 1 o feet above the present surface of the lake, showing that 
for a long period the lake stood at that level, he adopts no 
feet as the average level. This gives an average area of 550 
square miles, and a duration of 2,400 years. It seems certain, 
however, that the rock shelf dates from an earlier period, 
probably the close of the Quaternary pluvial period. Antevs 
(3, p. 102) quotes Gale and Huntington to the effect that there 
is a well-marked outflow channel through Emerson Pass, 
70 feet above Pyramid Lake. Jones (2, p. 40) gives the maxi- 
mum level of this pass as 78 feet, and states that there is no 
evidence of overflow, the summit of the pass having a floor of 
fine clays and silts. Jones realised the importance of this 
point, and appears to have examined the ground thoroughly. 
In either event, however, it is obviously impossible for 
Winnemucca and Pyramid Lakes to have stood more than 
78 feet above their present level without overflowing and 
becoming fresh. Hence the average level of 1 1 o feet above 
the present and the average area of 550 square miles are too 
great, and the age determination of 2,400 years too short. 
This method, therefore, gives the age of the present lakes as 
probably something less than 4,300 years. 

Thus we have four determinations, which are not, however, 
quite independent of each other. These give respectively 
3,880, 2,447, J >956, and 4,300 years. From these figures it 
seems probable that Pyramid and Winnemucca Lakes were 
fresh some time between 2,000 and 4,000 years ago. 

A lake may become fresh in one of two ways, either by 
overflowing into another basin, so that there is a flow of 
water through it which sweeps away the accumulated salt, 
or by becoming dry for a period long enough to allow the 
salts to* be buried below subaerial deposits. A short dry 
period will not suffice ; the overlying deposits must be so 
thick that when the lake forms again the salts are protected 


from the water and are not redissolved. The answer to the 
question in which of these two ways Pyramid and Winnemucca 
Lakes became dry, depends on the interpretation of the 
phenomena in Emerson Pass. If Jones is correct, the forma- 
tion of the present lakes was preceded by a long dry period 
which ended 2,000 to 4,000 years ago, and since then they have 
never overflowed. If Gale is correct, the lakes probably 
overflowed 2,000 to 4,000 years ago, and so became fresh, and 
there is no evidence of a preceding dry period. Since Jones 
examined the pass with Gale's work in mind, he is the more 
likely to be correct. 

This conclusion is supported by an investigation of W. van 
Winkle into the salt contents of Abert and Summer Lakes, in 
Oregon, which are remnants of the old Quaternary Lake 
Ghewaucan, north of Lake Lahontan, also without outlet. 
Van Winkle writes (3, p. 123) : " A conservative estimate of 
the age of Summer and Abert Lakes, based on their concen- 
tration and area, the composition of the influent waters, and 
the rate of evaporation heretofore assumed, is 4,000 years. 
It is quite possible that the lakes are recent pools, and that 
the salt and soda deposits of Early Quaternary Chewaucan 
Lake lie buried beneath them." 

It is known from the work of Antevs and de Geer that 
the major variations in the rate of recession of the ice-sheets 
at the close of the Quaternary glaciation in North America 
ran closely parallel with the variations in Scandinavia. In 
the post-glacial period, the peat-bogs show a succession of 
wet and dry periods which closely resembles the Scandinavian 
succession. H. P. Hansen (4, 5) states that in Eastern North 
America the climatic succession shown by the peat-bogs 
closely resembles that of North-west Europe : 

Eastern N. America. North-west Europe. 

Ice-retreat (Hudsonian) Late-glacial. 

Spruce, fir (cool, moist) Pre-boreal. 

Pine (warmer but still cool) Boreal. 

Oak and hemlock (warm, moist) Atlantic. 

Oak and hickory (warm, dry) Sub-boreal. 

Oak, chestnut, spruce (cooler, moister) Sub-atlantic. 


The pollen profiles reveal consistent and definite evidence 
for a dry period, which is best developed in east Washington 


and Oregon. In the North-west Pacific states of U.S.A. 
the dry period was less developed owing to the proximity of 
the Pacific. Early in the dry period there was a great 
explosive eruption of Mount Mazama ; the distribution of the 
pumice shows that at the time the winds blew from south and 
west. The climate was cooler and moister in the early 
post-glacial than at any time since, but the earliest forests 
differed little from the present. There is evidence of a rather 
wet period in the Puget Sound region some time alter 7000 
B.C. when the moisture-loving hemlock expanded. There 
was another abrupt spread of hemlock about 2000 B.C. (The 
dates appear to be estimates from the thickness of sedimentary 

K. Bryan (6) states that the bogs of Eastern Canada 
generally originated in ponds and lakes, bordered by a 
forest richer than the present, indicating a warm period. 
A cooler moister climate was followed by a warm dry period 
which again changed to the present cool moist climate. 
The changes in North America cannot be dated by archaeo- 
logical evidence, but there seems no reason to doubt the 
approximate synchronism of corresponding stages in North 
America and Scandinavia, especially as the two areas seem 
to be linked up to some extent by the deposits in Iceland 
and Greenland. This would give a long dry period during 
the third and second millennia B.C. in Eastern North America, 
which fits in excellently with the lake records. 

In this connexion there is some interest in a note in Nature, 
28th November 1925, p. 796, to the effect that the first 
settlement of the arid coast of Southern California is now 
dated about 3,000 years ago. The concordance with the lake 
evidence may be accidental, but it may mean that before that 
date the supply of fresh water was insufficient for settlement. 

The result of this discussion, therefore, seems to be that 
about 3,000 years ago there was either the end of a long dry 
period or a period of relatively heavy rainfall, probably the 
former, possibly both, and in any event an increase of rainfall. 
Jones' discussion of the levels indicated by the lake terraces 
is useless for the climates of the historical period, because 
they are all above the level of the Emerson Pass, and therefore 
belong to an earlier period in the history of Lake Lahontan, 
more than 3,000 years ago. 


Further evidence is supplied by the levels of Owens Lake, 
in Southern California (2, p. 200). There seems to be no 
doubt that the freshening of this lake occurred through the 
level rising so high during a wet period that the lake overflowed 
its basin. This lake is supplied by the Owens River, and, 
according to H. S. Gale, analyses show that the river, at the 
point where it was tested, would require 4,200 years to supply 
the chlorine and 3,500 years to supply the sodium now in 
Owens Lake. This gives 4,000 years as the maximum period 
since the freshening, but there are several factors which make 
this estimate too high. The analyses were taken at a point 
some way up the valley, and omit the lower third of the basin, 
which contains old saline clays, and across which the river 
flows slowly ; moreover, no allowance is made for greater 
rainfall in the past. Hence, Huntington concludes that the 
most probable length of the period which has elapsed since 
Owens Lake was fresh is between 2,000 and 2,500 years. 

Finally, we have the evidence of Walker Lake (2, p. 46). 
This is fed by Walker River, and there is no possibility that 
it was ever freshened by overflowing. Its salt content is 
only 0-25 per cent., and Jones calculates that the present 
rate of supply would accumulate this amount in about 1,160 
years or possibly less. Jones' theory of the origin of the lake 
is that it was formed by a change in the course of Walker 
River at the time of maximum level of Lake Lahontan, but 
this change, if it took place, must have occurred in Quaternary 
times. All that the lake shows us is that a minor dry period 
ended about 1,100 years ago. 

Thus we may sum up the evidence afforded by the lakes 
as follows : Some time after the great expansion of the lakes 
associated with the Quaternary glaciation of America, there 
occurred a long period of desiccation, in which Abert and 
Summer Lakes, and probably also Pyramid and Winnemucca 
Lakes, dried up completely. About 3,000 years ago this period 
of desiccation was brought to an end by an increase of rainfall 
to a value above its present amount, refilling the basins of 
Abert and Summer Lakes, but without causing them to 
overflow, filling the basins of Pyramid and Winnemucca 
Lakes, and perhaps causing them to overflow, and &lso filling 
Owens Lake and causing it to overflow. This was the time 
of greatest rainfall in Western North America during the 


historical period. Then followed a period of decreased rainfall, 
not enough to cause the complete disappearance of Abert, 
Summer, Pyramid, and Winnemucca Lakes, but enough to 
dry up Walker Lake for a period long enough to bury the salt 
accumulations. This secondary dry period ended about 1,100 
years ago, or A.D. 800. The lakes do not give any indication 
of the happenings after A.D. 800. There is some historical 
evidence ; thus Huntington (i) states that when the Aztecs 
founded the city of Mexico, about A.D. 1325, the level of the 
lake of Mexico was high, and that another period of high water 
occurred about 1550. 

Let us now see to what extent these lacustrine fluctuations 
fit in with the curve of tree growth. Unfortunately, the 
tree-growth curve helps us little during the early most crucial 
period of the change from dry to moist conditions. Antevs 5 
curves show a maximum at 840 B.C., a minimum at 740 B.C., 
and a second maximum at 660 B.C. Huntington's data 
show a maximum at 960 B.C., a minimum at 780 B.C., and a 
second maximum at 660 B.C. ; the maximum at 840 B.C. on 
Antevs' curves is barely indicated. These early growth 
curves are based on relatively few trees, and the corrections 
are uncertain, so that we cannot say more than that the wet 
period had definitely begun by 660 B.C., but may have begun 
two centuries or more earlier. It is interesting to note that 
the Chinese records also indicate a dry period from 842 to 771 
B.C. The rainfall maximum indicated by the overflow of 
Owens Lake presumably corresponds with the rapid growth 
of the trees from 480 to 250 B.C., shown on all the curves ; 
during this interval the rainfall reached its absolute maximum 
for the whole period since 1000 B.C. The period of drought 
during which Walker Lake dried up seems to extend from 
about A.D. 400 to 850 ; it is shown much more definitely 
on Huntington's than on Antevs' curve, owing to the different 
methods of correction adopted. The deep minimum shown 
by all the curves in the fifteenth century was apparently of 
too brief duration to cause even Walker Lake to dry up 

The alternation of dry and wet periods has also been traced 
by Hunlington (i) in the archaeological remains of Arizona 
and New Mexico. In these dry regions, whose crying need 
is water for agricultural purposes, he distinguishes three 


periods of maximum occupation or prosperity. In the oldest 
of these the people, whom he terms the Hohokam, were not 
limited to the neighbourhood of the water-courses but lived 
on the open plateau, and apparently depended on rainfall 
instead of on irrigation. The second people, the Pajaritans, 
lived partly on the irrigable land, but partly on the Pajaritan 
plateau. The latest pre-Columbian race was the Pueblo, 
who depended on irrigation, but lived in valleys where there 
is not now sufficient water for that purpose. The Pueblo 
village of Gran Quivera was still populous at the coming of the 
Spaniards. These three periods evidently correspond with 
three wet periods ; there is no evidence of continuity, and in 
some places, e.g., Ghaco Valley, the deposits containing 
remains of the different periods are separated by silts without 
human remains, formed during dry periods. The last of tJie 
three evidently represents the rainfall maximum from about 
A.D. 1 200 to 1400 ; it was the least important of the rainfall 
maxima. No dates can be assigned to the earlier periods, 
but the Pajaritan occupation presumably includes the period 
from 750 B.C. to A.D. 400, when the rainfall was much heavier. 
The first occupation, by the Hohokam, appears to be much 
older, and may have occurred during a very early rainfall 
maximum older than the oldest of the trees, but represented 
in Eastern North America by the period of peat-formation 
corresponding with the Atlantic stage in Europe. 

The lowest curve in Fig. 37 has been reconstructed from a 
paper by E. Schulman (7) in the Colorado Plateau, based not 
only on living trees but also on specimens of timber from ruins 
of Indian buildings. Apart from the trough and peak in the 
thirteenth to fourteenth centuries it tends to vary oppositely 
to Huntington's curve. This is probably due to its low 
latitude (35-40 N.) and is of interest in connexion with 
Huntington's theory of the shift of the climatic belts. With 
regard to the scale of the curves, it may be remarked that 
the maximum area of Pyramid and Winnemucca Lakes was 
less than one and one-half times the present area, so that the 
maximum of rainfall indicated by the curve is less than 50 
per cent., and probably not more than 25 per cent., greater 
than the present rainfall. 

Farther south we have the remarkable ancient Mayan 
civilisation of Yucatan (8, 9). This country is at present 


covered by almost impenetrable forests, the climate is hot, 
moist, and enervating, while the inhabitants are idle and 
uncultured. Buried in the forests are the ruins of great 
cities, decorated by elaborate carving, and indicating a greater 
and more progressive population and a high level of civilisation, 
one of the features of which, as is well known, was the con- 
struction of an elaborate calendar. The problems of Mayan 
history and chronology have not yet been completely solved, 
but it seems probable that before 400 B.C. there was little 
forest, the winters being dry and cool. Between 400 B.C. and 
100 B.C. the climate became somewhat moister and more 
uniform ; this is the time of the earliest carvings. The 
highest level of culture was reached in the period 100 B.C. to 
A.D. 300, first in the south, later in the north. By A.D. 300 
climate had become less favourable. The deterioration 
continued in A.D. 300-450 ; the forest advanced and civilisation 
declined in the south. From A.D. 450 to 900 the forest spread 
over the whole country, especially north Yucatan, and 
civilisation fell to a low ebb. There was a marked improve- 
ment in A.D. 900-1 100 accompanied by a great deal of building, 
but climate deteriorated again in 1100-1300. From 1300 to 
1450 there was a climatic improvement but culture did not 
respond to any extent. From 1450 onwards climate has been 
continually unfavourable. Sapper (9) thinks that the decline 
of civilisation was due partly to climatic changes and partly 
to the introduction of malaria. 

Since it takes time for both forests and civilisation to respond 
to the effects of climatic changes, we may date the latter about 
50 years earlier than the changes in the level of civilisation. 
This gives us the following comparison with Huntington's 
curve : 

Western U.S.A. Yucatan. 

Wet 500-250 B.C., IQO B.C.-A.D. 200. Dry 500 B.C.-A.D. 250. 

Dry A.D. 300- 800. Wet A.D. 400- 850. 

Wet A.D. 900-1100. Dry A.D. 850-1050. 

Dry A.D. 1100-1300. Wet A.D. 1050-1250. 

Wet A,D. 1300-1400. Dry A.D. 1250-1400. 

Dry A.D. 1450-1550. Wet A.D. 1400- 

In tha dry regions of Asia and Arizona the periods of high 
culture were attributed to an increase of rainfall, but Yucatan 
now suffers from too much rain, and any increase would make 


the conditions even less favourable than at present. Hence 
Huntington (8) infers that the great periods of Mayan history 
were times of decreased rainfall in Yucatan. Since they 
coincide with rainy periods farther north, we are evidently 
dealing here with a redistribution of rainfall. The way in 
which this was probably brought about will be discussed in the 
next chapter. 

The supposed climatic changes in Greenland have been 
a matter of controversy for many years, but excavations, 
described by Hovgaard (10), appear to establish their existence 
beyond doubt. Icelanders settled in Greenland in the tenth 
century A.D., and two colonies were established, the Eastern 
Settlement, just west of Gape Farewell, and the Western 
Settlement, 1 70 miles up the west coast. The settlers brought 
with them cattle and sheep, which were successfully reared ^at 
first, and they even attempted to grow grain, but before very 
long the colonies became dependent on supplies from Norway. 
Norway itself was passing through a time of stress, however, 
and the visits of ships became fewer and fewer, until some 
time in the fifteenth century they ceased altogether, and the 
colonies were lost sight of. For many centuries their fate was 
unknown, but the history of the Eastern Settlement has now 
been made out by the excavations of a Danish archaeological 
expedition at Herjolfsnes, near Cape Farewell. The most 
important evidence is derived from the excavation of the 
churchyard, in soil which is now frozen solid throughout the 
year, but which, when the bodies were buried, must have 
thawed for a time in summer, because the coffins, shrouds, 
and even the bodies were penetrated by the roots of plants. 
At first the ground thawed to a considerable depth, for the 
early coffins were buried comparatively deeply. After a time 
these early remains were permanently frozen in, and later 
burials lie nearer and nearer to the surface. Wood became 
too precious to use for coffins, and the bodies were wrapped 
in shrouds and laid directly in the soil. Finally, at least 
five hundred years ago, the ground became permanently 
frozen, and has remained in that condition ever since, thus 
preserving the bodies. The remains show a gradual deteriora- 
tion in the physique of the colonists ; their teeth specially 
are much worn, indicating that they lived mainly on hard 
and poorly nourishing vegetable food. 


The change of climate indicated by these facts is borne 
out by the evidence as to the ice conditions. When the 
colonies were first settled, there were traces of the former 
existence of the Eskimos, but none then lived so far south. 
The Eskimos follow the seals, which frequent the edge of 
the ice, and this indicates that in the tenth century the ice- 
edge in Baffin Bay lay far to the north. In the thirteenth 
century the Eskimos reappeared and advanced persistently 
southward, until by the middle of the fourteenth century 
they had occupied the Western Settlement, which apparently 
they destroyed. 

The accounts of the early Norse voyages to Greenland 
are remarkably free from references to ice conditions, and, 
in fact, as O. Pettersson (n) points out, it is difficult to 
uaderstand how their protracted explorations could have 
been carried out if the ice conditions had been anything 
like those of the present day. Pettersson's chart of the old 
Norse sailing routes shows a track direct from Iceland to the 
east coast of Greenland in latitude 66 N., then down the 
coast to Cape Farewell, and up the west coast. According 
to the documentary evidence which he adduces, this route- 
at present almost impossible was followed until about 
A.D. 1 200, when it was abandoned for a more southerly route. 
On the other hand, as early as A.D. 998 a shipwrecked party 
was ice-bound on the east coast of Greenland, probably near 
or north of Angmagsalik. It is to be noticed that the ship was 
wrecked on the coast and not on the ice. 

The early climatic history of Greenland, therefore, appears 
to have been somewhat as follows : When the country was 
colonised in the tenth century its climate was much more 
favourable than at present, for herds of sheep and cattle 
thrived. There was less ice than at present in the East 
Greenland Current, and it is even possible that at first there 
was no ice at all ; Baffin Bay seems to have been largely free 
of ice. But in the second half of this century the climate was 
already deteriorating, and about A.D. 1000 there came a 
foretaste of the coming ice. After this, conditions apparently 
improved slightly, and the colony appears to have prospered 
during oaost of the eleventh and twelfth centuries. Towards 
the close of the twelfth century deterioration again set in, 
and the ice conditions rapidly became very bad. The 


summer thaw became shorter and shorter, and about A.D. 
1400 the ground became permanently frozen. Communica- 
tion with the mother-country was broken, life became too 
hard to bear, and the colonies finally perished. 

For South America the only evidence I can find of climatic 
changes in the historical period is given by E. Taulis (12) 
who estimated the rainfall of each year from 1535 to 1931 in 
Chile on a scale of 1-5. From his figures it appears that there 
were wet periods about 1550 and in 1684 to 1700, and a great 
drought from 1770 to 1783. This agrees with Antevs' curve 
of tree growth. 


(1) HUNTINGTON, E. " The climatic factor, as illustrated in arid America." 

Washington, 1914. 

(2) WASHINGTON, CARNEGIE INSTITUTION. Publication No. 352. " Quaternary 

climates." Papers by J. CLAUDE JONES, ERNST ANTEVS, and ELLSWORTH 
HUNTINGTON. Washington, July 1925. 

(3) WASHINGTON, U.S. GEOLOGICAL SURVEY. Water Supply Paper 363. 

" Quality of the surface waters of Oregon." By WALTON VAN WINKLE. 

(4) HANSEN, H. P. " Postglacial forest succession and climate in the Oregon 

Cascades." Amer. J. Sci., 244, 1946, p. 710. 

(efi > " Postglacial forest succession, climate and chronology in the 

Pacific North-west." Philadelphia, Trans. Amer. Phil. Soc., 37, Pt. i, 1947. 

(6) BRYAN, K. ** Palaeoclirnatology in North America as a result of the study 

of peat bogs." <X Gletscherk., 20, 1932, p. 76. 

(7) SCHULMAN, E. " Nineteen centuries of rainfall history in the Southwest." 

Milton, Mass., Bull. Amer. meteor. Soc^ 19, 1938, p. 311. 

(8) HUNTINGTON, E. "Civilisation and climate." 3rd ed. New Haven, 1924. 

(9) SAPPER, K. " Klimaanderungen und das alte Mayareich." Beitr. Geoph., 

Leipzig, 34 (Koppenbd 3), 1931, p. 333. 
(10) HOVGAARD, W. " The Norsemen in Greenland. Recent discoveries at 

Herjolfsnes." New York, N.Y., Geogr. Reu., 15, 1925, p. 605. 
(u) PETTERSSON, O. "Climatic variations in historic and prehistoric time." 

Svenska Hydrogr.-Biol. Komrn. Skr., 5. Goteborg, 1914. 
(12) TAULIS, E. " De la distribution des pluies au Chili. La periodicit6 des 

pluies depuis quatre cents ans." Geneve, Mat. e'tude calam., 9, 1934, p. 3. 



FROM the preceding four chapters we see that during 
the historical period there have been several climatic 
fluctuations of quite appreciable magnitude ; in some 
parts of the Northern Hemisphere the fluctuations were closely 
similar over wide areas, while in other parts they were in 
distinct opposition. We must now try to discover the causes 
of these variations. Fig. 38 gives general curves of estimated 



&C, Q A.CX 



Fig. 38. Variations of rainfall, world. 

rainfall for Europe, Asia, North America and for East Africa 
from the equator northwards. The curve for Europe was 
constructed by superposing the first three curves of Fig. 31 
and drawing a mean curve through them ; that for U.S.A. was 
drawn in the same way from Huntington's and Antevs' curves. 
The first three curves of Fig. 38 show a good deal of resemblance 
with some discrepancies which may be due to difficulties of 
precise dating. The discrepancies in the curve for Africa are 
greater but, as previously stated, the dating of this curve is 
only conjectural ; the maximum shown at 1250 B.C. may 
easily be pushed back to 2000 B.C. 



The first three curves all refer to the region between about 
35 and 65 N., in which the rainfall is mainly brought by 
barometric depressions (Chapter II.), while the curve for 
Africa aims at showing the variations in the equatorial belt 
of low pressure. Between these two lies the sub-tropical high 
pressure belt where the rainfall is mainly monsoonal, and is 
greatest when the general circulation of the atmosphere is 
weakest. Our information about past climates in this region 
is scanty, but such as it is, it suggests that the variations were 
in the opposite direction to those farther north. First we have 
the rainfall maximum at Mohenjo-Daru in India about 
2750 to 2500 B.C., which comes in the middle of the long dry 
period in Europe and Asia. Then the variations in Yucatan 
are directly opposed to those in the Western United States, 
and there are indications that in the later stages at least the 
variations of rainfall in Cambodia agreed with those in 
Yucatan. This zonal distribution strongly suggests that the 
variations of rainfall are related to changes in the zonal 
circulation of the atmosphere. 

Since the total amount of air is fixed, the average barometric 
pressure over the whole surface of the planet must be always 
the same, and an excess in one region must be compensated 
by a deficit in some other region. Now it has been found, 
especially by Sir Gilbert Walker (i), that this process of 
compensation is not haphazard ; it follows a clearly marked, 
though not inviolable rule. When, in a region where pressure 
is normally high, such as one of the sub-tropical anticyclones, 
it rises even higher than usual, there is a tendency for pressure 
to be higher than usual in all those parts of the world where it is 
normally high, and lower than usual in all those parts of the 
world where it is normally low. That is to say, if pressure 
is above normal in, for example, the Azores anticyclone, it 
will tend to be above normal over a belt stretching more or 
less completely round the globe from Hawaii to the north of 
Mexico, and across Bermuda and the Azores to North Africa. 
On the other hand, pressure will tend to be below normal 
over the belt of storminess which runs from Kamchatka and 
the Aleutian Islands across Southern Canada and New- 
foundland to Iceland, the British Isles, Norway, a*nd the 
North of Asia. At the same time there will also be a tendency 
for pressure to be below normal near the equator. 



Some possible causes of these variations of post-glacial 
climate are shown in Fig. 39. This is divided at 500 B.C. into 
two parts, the time-scale on the right being two and a half 
times that on the left. First we have variations of sea level. 
These are known accurately only in Scandinavia, but this 
region is important because of its proximity to the only broad 
gateway to the Arctic. A little before 5000 B.C. the Ancylus 
emergence gave place to the Litorina subsidence, which 
reached its maximum between 4500 and 4000 B.C., after which 
the sea gradually receded. By 500 B.C. it had reached nearly 
its present level, and since then the changes have been un- 
important. It was shown in Chapter VIII. that the decrease 
of continentality at the maximum of the Litorina Sea must 
have raised the winter temperature by about 5 F., and that 
this agreed closely with the observed rise of temperature <dn 
Scandinavia. The subsidence of the land, and the readier 
access of southerly winds, would also affect conditions in the 
Arctic Ocean, decreasing the area of floating ice, and so 
probably cause a general amelioration of temperature over 
all the higher latitudes of the Northern Hemisphere. 

Evidence of a post-glacial Climatic Optimum has been 
found in Franz Josef Land, Spitsbergen, Norway, the Baltic 
Shores, Iceland, Greenland, Ireland, Eastern and Central 
North America, Patagonia and Tierra del Fucgo, New Zealand, 
Southern and Eastern Australia, South Africa and the Ant- 
arctic. We do not know that all these are of the same date, 
but there is one feature common to a large proportion of the 
deposits which points strongly in this direction the bulk of 
the evidence for the post-glacial Climatic Optimum is 
derived from or associated with beaches raised a few feet 
above the present sea-level. In the Baltic the raised beaches 
are higher and are definitely associated with a subsidence of 
the land, but in most parts of the world the change of level 
was remarkably uniform. A uniform change of level at many 
far distant points is almost certainly due to a rise of the sea 
and not to a subsidence of the land. A rise of sea-level may 
be due to one of three causes : 

(a) A decrease in depth of part of the sea floor, com- 
pensated by a decrease in the elevation of part of 
the land area. 


(b) An increase in the volume of sea water without 
change of mass, owing to a decrease in density. 

(c) The actual addition of water to the oceans. 

The general rise of sea-level during the Climatic Optimum 
is hard to estimate precisely, owing to the difficulty of obtaining 
the exact levels of old sea beaches, but it seems to have been 
of the order of 10 feet. Of the three possible causes of this 
rise, (a) can be dismissed very shortly. The principal land 
areas in which there was extensive post-glacial subsidence 
are Scandinavia and North America north of the Great 
Lakes, but both these subsidences were largely compensated 
by elevation of the land to the southward ; moreover, the 
period of maximum depression was probably over before the 
height of the Climatic. Optimum. Causes (b) and (c) are 
both possible, but are difficult to estimate. 

The mean depth of the oceans is approximately 12,000 feet. 
Taking the coefficient of expansion of water as -00015 f r 
one centigrade degree, we find that an increase of temperature 
by i C. or 1-8 F. would raise the mean level of the surface 
by i 8 feet. Thus a rise of the mean temperature of the whole 
mass of the oceans by 5 F. would raise the general level by 
5 feet. The increase in the surface temperature of the 
northern North Atlantic approached 5 F. and in the Arctic 
and Baffin Bay the increase was probably even greater. The 
temperature of the lower oceanic layers is determined by that 
of the polar oceans, but the warming of the whole ocean mass 
would be very slow and the general rise of temperature is not 
likely to have been nearly so much. 

The only ways in which water can be added to the ocean 
are by a decrease in the level of enclosed lakes unconnected 
with the sea, and a decrease in the volume of the ice-sheets 
and glaciers. The volume of water in enclosed lakes without 
outlet is so small in comparison with the area of the oceans 
that it can be neglected. The ice-sheets, however, are on a 
different scale. The ice-covered area in Greenland and the 
Antarctic is about six million square miles, and the average 
thickness of the ice is nearly 5,000 feet. If all this ice were 
melted^ it would raise the general level of the oceans by from 
140 to 190 feet. The area occupied by the oceans is about 
140 million square miles, or 23 times the area occupied by 


ice, so that in order to raise the level by ten feet, it would be 
necessary to melt off 230 feet of ice. Because of the difference 
of density between glacier ice and water we may put the figure 
at 250 feet. Now we know that even in the much less intense 
warm period of the early Middle Ages the boundaries of the 
Greenland ice-sheet retreated appreciably, which implies a 
corresponding diminution of thickness, so that in the pro- 
longed warm period of the Climatic Optimum a lowering 
of the average level of the ice-sheets by 250 feet is quite possible. 
These two factors, increase of ocean temperature and increase 
in the mass of water, appear to be quite competent between 
them to raise the general level of the oceans by 10 feet, the 
greater part of this being due to the melting of ice. 

The second factor to be considered in our climatic re- 
construction is the annual range of temperature. As described 
in Chapter V., the obliquity of the ecliptic appears to have 
reached a maximum about 8150 B.C., and to have decreased 
steadily since that date. Also, about 8500 B.C. the earth 
was farthest from the sun (aphelion) in the northern winter, 
whereas it is now farthest from the sun in the northern summer. 
Both these factors would cause an appreciably greater seasonal 
range of radiation in the ninth millennium B.C. than at present. 
This change is shown by the second full curve on the left of 
Fig- 39- 

Over most of the world these " astronomical " changes do 
not affect the total supply of solar radiation appreciably, 
but in the Arctic, which receives little or no solar radiation 
in winter, the effect of increased seasonal contrast is to increase 
the total solar radiation considerably. 

The broken curve between these two full curves represents 
the variation of temperature in North-west Europe, copied 
from Fig. 32. It is seen that the left-hand part of this curve 
is a mean between the curve of sea-level and that of summer 
radiation. The " Climatic Optimum " occurred about 5000 
B.C., after which temperature fell gradually until about 
3000 B.C. The fluctuations in the years between 3000 B.C. 
and 500 B.C. will be discussed later. 

The climate of the Boreal phase, about 6000 B.C., appears 
to have been definitely more continental than at g/iy sub- 
sequent time, with cold winters and hot dry summers in many 
parts of the Northern Hemisphere, This is probably due to the 


combination of high obliquity with winter in aphelion ; in 
North-west Europe the larger land area and the shutting off 
of the Baltic was also a factor. The present situation, 
decreased obliquity and winter in perihelion, gives the 
opposite effect of mild winters and cool summers. The 
ice-sheets retreated rapidly between 8000 and 7000 B.C., 
while since the beginning of the Christian era there does not 
seem to have been any general retreat, only long-period 
oscillations about a mean position. To this extent the long- 
period changes of climate since 8000 B.C. support the 
astronomical theory. 

The third factor is the variation of solar activity. Sir 
Gilbert Walker (2) has pointed out that the contrast between 
the zones of low and high pressure is apparently controlled 
tQ some extent by variations of solar activity. When sunspots 
become more numerous, pressure increases in the areas where 
it is already high, and decreases in those where it is already 
low. In the temperate storm belts, a high sunspot number 
tends to be associated with low pressure, great storminess, and 
heavy rainfall. According to Huntington and Visher (3), 
the belt of storminess in the Northern Hemisphere moves 
southward and increases in intensity at times of many sun- 
spots, but moves northward and decreases in intensity at 
times of few sunspots. C. E. P. Brooks (4) found that the 
annual frequency of thunderstorms shows a fairly close 
relation to the sunspot number. In many parts of the world, 
including Siberia, Sweden, Norway and Scotland in the 
north and the West Indies, South-eastern U.S.A., Southern 
Asia and the Tropical Pacific in the south, the frequency of 
thunderstorms is greatest when sunspots are most numerous. 
Between these two belts is a region including England and 
Wales, Holland, Germany and the Northern and Western 
U.S.A., in which the relation is small, but still generally 
positive. Since in the interior of the continents and in 
tropical regions a good deal of rain is associated with thunder- 
storms, this suggests that rainfall maxima should coincide 
with maxima of sunspots, and there is other evidence that on 
the whole the total rainfall over the land areas is greatest 
when sunspots are most numerous. Our next step must, 
therefore, be to construct a curve which will represent the 
variations of solar activity over as long a period as possible. 


For our knowledge of sunspot frequencies since 1749 we 
are mainly indebted to the researches of R. Wolf, who has 
compiled a complete table beginning with that year (5). 
The earlier data are based mainly on a long but rather 
fragmentary series of records from China. The first record 
of a sunspot occurs in the Chinese archives in A.D. 188, and 
the first aurora in A.D. 194, but records only become frequent 
in the fourth century, apparently reaching a maximum about 
374. Another maximum, both of spots and aurora, occurs in 
535-540, the first half of the sixth century giving us 20 records 
of spots and 13 of aurorae. In the seventh and eighth centuries 
the number of records is very small, rising to another maximum 
about 840, in which year there are records of 90 sunspots, 
while brilliant aurorae occurred in 839 and 840. The records 
of sunspots again become very few between A.D. 850 and 107,0, 
though there is a secondary maximum of aurorae about 993. 
There is a great outburst of sunspots in the years 1077 to 1079, 
and the frequency of both spots and aurorae remains very 
high until about 1250, with probably a secondary maximum 
about 1 20 1. The last half of the thirteenth and the first 
half of the fourteenth centuries again show a falling off in 
the records, but about 1370-1375 they become very numerous, 
and WoLer considers that the absolute maximum of solar 
activity during the Christian era occurred in 1372. If so, 
this maximum was of very brief duration, tor there are no 
records of sunspots between 1383 and 1511, while the frequency 
of aurorae also decreases. From about 1676 to 1725 there 
was an extraordinary dearth of sunspots, followed by maxima 
about 1778 and 1837. 

From these figures of sunspots and aurorae an attempt 
has been made to construct a curve of solar activity since 
the occurrence of the first spot in A.D. 188. The early portion 
of this curve is not reliable, probably depending more on the 
accidental circumstances which led to the making and pre- 
serving of records than on the variations of the phenomena 
observed, but it seems probable that the maxima of the 
eleventh and fourteenth centuries, and perhaps also that 
of the ninth century, are real. From 1750 to 1940, the curve 
is based on xo-year means of the relative numbers. This 
curve of solar activity is shown at the top right-hand side of 


The construction of a curve of solar activity during the 
past few thousand years would be facilitated if we had any 
definite knowledge as to the cause of the sunspot cycle. 
Various hypotheses have been put forward, the favourite 
being the disturbance of the sun's surface by the influence 
of the planets, especially Jupiter. Jupiter completes his 
journey round the sun in 1 1 -86 years, but when the influence 
of the other planets is added to that of Jupiter the result is an 
irregular recurrence with an average length of slightly less than 
1 1 8 years, which bears some resemblance to the sunspot 
curve. The divergences are, however, too great for the com- 
plete acceptance of this theory. H. H. Turner (6) has 
devised an interesting alternative, which supposes that sun- 
spots are due to the impact of meteors belonging to a swarm 
(tjie " Sunspot Swarm ") which pursues an elliptical orbit 
round the sun with a period which averages slightly over 
eleven years, but varies in length owing to interference with 
the Leonid Swarm. The latter has a period of 33^ years. 
According to Turner, the " Sunspot Swarm " originated in 
A.D. 271 owing to the Leonid Swarm coming into conflict 
with the rings of Saturn, but so far as I am aware there is 
very little if any positive evidence for the existence of the 
Sunspot Swarm of meteorites, and in Chapter IV. we found 
some evidence for the existence of an eleven-year cycle in 
meteorological phenomena long before the beginning of the 
Christian era. 

The two curves on the right below the sunspot curve show 
the variations in the thickness of the annual layers of Lake 
Saki, South Russia, and the thicknesses of growth rings of 
trees in Western U.S.A. according to Antevs, slightly smoothed. 
Previous to about A.D. 800 these do not show much relation- 
ship to the curve of solar activity, but as previously stated, 
the latter curve is not reliable for this period. The minimum 
about A.D. 250, for example, may be due entirely to a gap in 
the records. The long minimum of solar activity between 
600 and 750 is reflected in the Russian curve, as is the peak 
about 850. The sunspot maximum of 1077-1079 is reflected 
in the highest peak of Antevs' curve, though the latter appears 
to come^ about 30 years earlier, and the general shape of the 
Russian curve from noo to 1300 is very similar to the solar 
curve. The great peak about 1372 also appears in both 


rainfall curves, though again somewhat early in the tree rings. 
Finally, the long dearth of sunspots from 1676 to 1725 is faith- 
fully reflected in the annual layers ; it is also shown in the 
actual rainfall observations in Western Europe (Chapter 


The short record of the low-level stage of the Nile (Fig. 35) 
can only be relied upon between A.D. 640 and 1400, but 
during this period it presents considerable similarity to the 
unspot curve. Thus we have : 

Sunspot maxima A.D. 620 840 1077 I2O I 37 
Nile, low-level stage . 645 880 uoo 1225 X 375 

The maxima of level in the low-water stage of the Nile 
apparently follow sunspot maxima by intervals of from five 
to forty years, but the order of importance of the maxin>a 
differs greatly in the two curves. In the Nile, the great 
crest at uoo completely dominates all the later variations,, 
and the peak at 1375 is insignificant. In this connexion it 
must be remembered that the Nile curves have been corrected 
on the assumption that the deposition of alluvium raises the 
level of the whole valley, including the low-level channel, 
at a uniform rate. The alluvium is deposited by the flood ; 
the low stage of the Nile is supplied by water which has been 
filtered by its passage through a series of lakes. Hence a 
consistently high level at the time of low water, as happened 
about uoo, might perform a good deal of erosion, and by 
cutting out a deep if narrow channel, result in a long series 
of very low minimum levels in subsequent years. This 
would be facilitated by the series of weak floods, which 
apparently occurred at about the same time, and which 
would bring little alluvium. For these reasons it is possible 
that the depression between about A.D. 1150 and 1400 may 
not indicate the true level of equatorial rainfall. 

Finally we come to a hypothesis due to O. Pettersson (7) 
that variations of climate in the historical period have been 
caused by long-period variations in the circulation of the 
oceans caused by changes in the " tide-generating force. 55 
The latter varies with the declination and proximity of the 
sun and moon to the earth and, in addition to shorter variations, 
reaches maxima at variable intervals which average about 
1,700 years. Pettersson gives the dates of these maxima as 


about 3500 B.C., 1900 B.C., 250 B.C., and A.D. 1433, and of 
the minima as about 2800 B.C., 1200 B.C., and A.D. 550. These 
variations are shown in the middle curve of Fig, 39. 

Pettersson points out that in addition to the surface tides, 
which would have a greater range at maxima than at minima 
of the tidal force, there are also internal tides, formed where 
a relatively light less saline layer rests on a heavier more 
saline layer, and these submarine tides have actually been 
measured in the entrances to the Baltic, attaining a range 
of 80 to 90 feet. Submarine waves also enter the Arctic basin, 
where they were first traced by Nansen. At periods of 
maximum tide they are stronger and are able to break up the 
ice, leading to an increase in the amount of drift ice carried 
out into the North Atlantic by the polar currents. At tidal 
minima, on the other hand, the ice is broken up to a much 
less extent, and so there is little drift ice in the Greenland and 
Iceland seas. 

Drift ice is an important factor in increasing the storminess 
and consequently the rainfall of temperate latitudes, and in 
deflecting the storm tracks into lower latitudes, and we should 
accordingly expect the maxima of tidal force to be maxima 
of rainfall also. This appears to be the case with the tidal 
maxima of 1900 B.C., 250 B.C., and A.D. 1433, while the minima 
in 2800 B.C., 1 200 B.C., and A.D. 550 are also clearly shown. 
Further, with increased drift ice in the Atlantic we should 
expect lower temperatures in the coastal regions of Western 
Europe, and the temperature curve shows that on the whole 
this was so. It seems that there is good support for 
Pettersson's theory as well as for that of solar activity, and that 
the actual variations of climate since about 3000 B.C. may 
have been to a large extent the result of these two agents. 

To facilitate comparison, the rainfall curves for Europe, 
Asia, and the U.S.A. in Fig. 38 have been combined in the 
lowest full curve of Fig. 39. This was constructed by first 
superposing the three curves and sketching in a general 
average. The positions of maxima and minima on all the 
individual curves of Figs. 31, 33, 35, and 36 were then marked 
on the curve, which was adjusted to bring the peaks and troughs 
into accord with the " majority verdict." This was done to 
eliminate as far as possible the uncertainties of dating. The 
broken curve is Pettersson's curve, extended backwards by 



assuming a periodicity of 1,750 years, and modified, after 
A.D. 100, by superposing on it the curve of solar activity. 

The results are of great interest. From A.D. 100 onwards the 
fit is quite as good as can be expected from the nature of the 
data, both in the broad swing and in the peaks at intervals of 
two or three hundred years. From 3000 B.C. to A.D. o the long- 
period variations of rainfall also fit very well. From 2000 
B.C. onwards, shorter oscillations of rainfall are superposed on 
these long waves, and it is a reasonable supposition that these 
may also be related to variations of solar activity. Before 
3000 B.C., however, the two curves are in direct opposition. 
There are three possible reasons for this : 

1. The dating of the rainfall curve is incorrect. Although the 
curves for Europe and Asia were constructed independently, 
their dating before 2500 B.C. ultimately depends mainly 
on the dating of the cultures of Egypt and the Euphrates 
valley, which may not yet be quite established. 

2. The extrapolation backwards of the tidal curve may be 
incorrect. The error is unlikely to be sufficiently great 
to invalidate the apparent opposition. 

3. The climatic effect of tidal variations was reversed about 
3000 B.C. The possibility of this depends on the ice 
conditions in the Arctic. The apparent opposition may 
be accidental, due to the fact that the Litorina subsidence 
happened to coincide with a minimum of the tidal force. 
It is of interest, however, to attempt a reconstruction of the 
history of the Arctic on the assumption that the opposition 
is real. 

During the Climatic Optimum the mean temperature of 
the Arctic was many degrees higher than now, as shown, for 
example, by the growth of peat-bogs in Spitsbergen. Owing 
to the high obliquity and winter in aphelion the winters were 
cold and the summers very mild, the net result being a gain 
of solar heat. The Arctic Ocean may have been in a stage 
at which ice floes tended to form in winter but to break up 
and melt in summer without much ice finding its way into the 
polar currents. In such conditions a slight excess or deficit 
of heat would make a large difference to the development of the 


In periods of minimum tidal force the amount of warm 
water finding its way into the Arctic basin was also a minimum, 
and an ice-sheet would be likely to form in winter, which, 
when it broke up in summer, would supply some ice for the 
polar currents to carry into the Atlantic, though less than at 
present. These were the rainy periods. At maximum tidal 
force on the other hand the Arctic received a large amount of 
warm saline water. Both the high temperature and salinity 
would act against the freezing of the ocean surface, so that 
in these periods there may have been either no ice at all or so 
little that it broke up and melted away in summer without 
any ice reaching the North Atlantic. In such circumstances 
depressions would tend to follow northerly tracks into the 
Arctic Ocean instead of across Europe, giving a dry period in 
Western Europe. This is not entirely speculative ; much the 
same happens under present conditions when a long spell of 
southerly winds between Iceland and Novaya Zemlya drives 
the ice edge unusually far north, and this is almost invariably 
followed by a drought in Western Europe (8). Tidal force 
alone, however, cannot account for the relatively heavy 
rainfall of the Atlantic period compared with the present. 
In Europe this might be put down to the larger and warmer 
Baltic of the Litorina Sea, but if as seems probable the rainy 
period occurred also in Asia and North America some more 
general cause must be sought. This may be either the 
generally higher temperature of the oceans due to the smaller 
amount of sea and glacier ice, or possibly a period of increased 
solar activity. 

By about 2500 B.C. these favourable conditions had largely 
passed. Scandinavia had risen almost to its present level, 
and the contrast between winter and summer had greatly 
decreased. Hence the Arctic Ocean became cooler and more 
liable to freeze. The main characteristic of the Sub-boreal 
in Western Europe seems to have been the instability of its 
climate, periods of drought and heat alternating with periods 
when the climate resembled the present, at intervals of perhaps 
a few hundred years. It is not unlikely that the Arctic ice-cap 
had now reached the critical stage between non-persistence 
and persistence it was difficult for it to become established, 
but once firmly formed it was difficult to destroy. In such 
circumstances wide oscillations of climate would be expected. 


I think that, paradoxically, persistence would be aided by 
increased tidal force, which would cool the Gulf Stream Drift 
by the ice carried into it by the polar currents. It must be 
remembered that a firm unbroken ice-cap grows more slowly 
than broken sea ice which allows the sea to freeze between 
the ice-floes. 

The final stage came about 500 B.C. when for some reason 
the Arctic ice-cap at last became firmly established, apparently 
very extensively, after a few centuries of heat and drought. 
The reason for this change is not clear ; it may have been 
due to a change of solar activity or possibly to explosive 
volcanic activity. This period of an established Arctic ice-cap 
and stormy weather apparently lasted for about a thousand 
years, but the favourable climate of Iceland and Greenland, 
and the absence of ice in the accounts of the early Nopse 
voyages, suggest that some time after A.D. 500 the ice-cap 
reverted to the semi-permanent stage, and remained so unti] 
nearly 1200. 

The latest maximum of tide-generating force is dated by 
Pettersson as A.D. 1433. Although this force has been repre- 
sented in Fig. 39 by a smooth curve, this is far from being 
the real condition. Superposed on the long period are shorter 
ones of about 90 and 9 years. From Pettersson's diagram 
the actual maxima apparently occurred in three sharp peaks 
about 1340, 1430, and 1520. Now the main peak on Antevs' 
curve of tree-growth comes in the decade 1331-1340, with a 
secondary peak about 1431-1440. In the Lake Saki varves the 
peak is about 1450 with secondary peaks at 1390 and 1530. 
These dates fit in rather better with Pettersson's curve than with 
the curve of solar activity. We note also that about 1340 
the westerly winds in England were more persistent than at 
present, indicating a more stable Icelandic low. 

On the other hand the great "storm floods' 5 of the twelfth to 
fourteenth centuries, on which Pettersson sets great store, come 
on the whole before the absolute maximum of Pettersson's 
curve. The maximum damage occurred in 1170-1178, 1240- 
1253, 1267-1292, 1374-1377, and 1393-1404, the middle one 
being the most prolonged and severe ; there was also a great 
flood in 1421. These fit in better with the sunspot cprve than 
with Pettersson's. Marine inundations require a combination 
of violent storms and high tides and these disasters of the 


twelfth to fourteenth centuries may represent maximum stormi- 
ness in the North Sea associated with great solar activity, at a 
time when the tidal range was approaching its maximum. 
There may also have been a slight subsidence of the land about 
this time, especially in the fen districts where drainage opera- 
tions had been carried on. In any case the main period of 
great marine inundations would not be likely to continue after 
the tidal maximum, for by that time the most vulnerable areas 
would have been overflowed and protective measures taken. 

The remarkable agreement between Pettersson's rather 
obscure " tide-generating force " and the major variations 
of climate since 3000 B.C. is surprising, and seems to show that, 
under favourable conditions, comparatively small causes may 
have disproportionately large effects. The favourable con- 
ditions are the stratification of the upper layers of the North 
Atlantic, the existence of the submarine Wyville Thomson 
ridge at just the right depth between Atlantic and Arctic, 
the critical stage of the Arctic ice, just over the border between 
non-glacial and glacial, and the maximum range of the tidal 
force itself. Such a coincidence is not likely to have recurred 
often in geological time, and in spite of the apparent effective- 
ness of Pettersson's tidal force in recent millennia we may 
safely discount it as a permanent agent in climatic changes. 

We may conclude this summary of post-glacial history with 
brief references to four more recent climatic chapters : 

1. The dry period of the sixteenth century. 

2. The great outburst of glaciers about 1600. 

3. The dry period of 1701-1750 in Western Europe. 

4. The rise of winter temperature in 1850-1940. 

Comparatively rapid variations of climate, of the order of a 
century, have presumably always occurred, and are shown 
in the thicknesses of the glacial varves, lake deposits and tree 
rings, but they can be properly examined only when we can 
assemble sufficient facts, especially about the prevailing winds, 
to enable us to reconstruct the probable pressure distribution, 
and this is not possible before the sixteenth century. Here we 
have to add another factor to our list of causes, namely, the 
variations of the atmospheric circulation referred to on 
p. 66. The atmosphere, like the sea, is in a state of perpetual 


oscillation, the " waves " varying in length from a few hours 
to many years, the result being highly complex changes in the 
distribution of pressure from day to day, month to month and 
year to year. These changes can be predicted for a day or 
two, and are the basis of modern weather forecasting. Much 
effort has been devoted to their analysis in the hope of fore- 
casting for longer periods ahead, but they are only periodic 
to a slight extent and so far the attempts have been un- 
successful. It is highly probable, however, that the longer 
oscillations are due partly to variations of solar activity and 
partly to interactions between the circulations of the atmosphere 
and the oceans. They take the form of an alternate weakening 
and strengthening of the whole circulation of the atmosphere. 
In the periods of weak circulation the low pressure centres 
near Iceland and the Aleutians are smaller, shallower a$d 
less stable, and anticyclones readily develop over the western 
margins of the continents. The winds are variable and thq 
climate is " continental," with cold dry winters and hot 
summers. In the periods of strong circulation the low 
pressure areas are enlarged and intensified and powerful 
south-westerly air streams invade the western parts of the 
continents. The climate becomes " oceanic " with mild 
rainy winters and cool summers. Even in the middle of an 
" oceanic " period, however, there are occasional " con- 
tinental " years, such as 1921, and vice versa, and the change 
from one type to the other often seems to be abrupt. 

i . The latter half of the sixteenth century appears to have 
been mainly " continental," rather dry on the whole in the 
north temperate zone. In Western Europe the winds were 
probably more easterly than now, and the winters were cold. 
This seems to have been a time of minimum solar activity. 
There are few records of storms, and a reasonable inference is 
that the floating ice-cap suffered little disturbance and was able 
to grow in extent and solidity. The result would be more 
frequent incursions of cold Arctic air over Russia and extensions 
of the Siberian anticyclone across Northern Europe. This 
would give Northern and Western Europe frequent easterly 
winds, cold in winter, hot in summer. Similar conditions 
probably occurred in North America. The mayi storm 
tracks were deflected southwards, and this period seems to 
have been rainy both in South-east Europe and in Yucatan. 


2. The great outburst of mountain glaciation which began 
at the end of the sixteenth or early in the seventeenth century 
was so remarkable that this period has been termed the " Little 
Ice- Age." In the Alps and Iceland it began about 1600 
and reached a maximum about 1643. In both countries the 
advances exceeded those at any other period since late-glacial 
times. There was a retreat in the first half of the eighteenth 
century, followed by a readvance in the first half of the 
nineteenth century, which gave place to a rapid retreat after 
1850. In Southern Norway and Alaska, on the other hand, 
the maximum advance did not occur until about 1750. 

It is now generally agreed that the most favourable con- 
ditions for the growth of glaciers are snowy winters and cool 
damp summers. The snowfall, however, takes time to 
accumulate, and the maximum extension of a glacier lags 
behind the greatest accumulation of snow by a number of 
years, depending on the size and length of the glacier. Hence 
the " glacial period " which began about 1600 probably 
reflects the snowfall of the latter part of the sixteenth century. 
We have seen that this period was probably mainly anti- 
cyclonic over Northern Europe, and that depressions 
followed southerly tracks. The greatest snowfall occurs in 
the northern halves of depressions, consequently this was a 
time of heavy snowfall in the Alps, Pyrenees and Iceland. 
During this period the snowfall accumulated at high levels, 
but at first the glaciers were unable to extend into the valleys 
because of the low mean annual temperature, which decreased 
their viscosity and kept them frozen to the ground. As soon 
as this limitation was removed, the glaciers grew rapidly in 
extent. At this time, however, Norway and Alaska still 
remained under the influence of the northern anticyclone, 
and the glacial advance did not extend to those regions until 

3. The first half of the eighteenth century is the first period 
for which instrumental observations are available. This period 
was discussed by G. E. P. Brooks (9). The greater frequency 
of north-easterly winds and the light rainfall over Western 
Europe (see p. 309) point to a decreased intensity of the 
Icelandic low, and there is some suggestion that the Aleutian 
low was also weak. In summer, anticyclones tended to 
develop over Western Europe. The lack of snowfall and 


probable high summer temperatures caused a recession of 
the Alpine glaciers, but in Norway snowfall increased and the 
glaciers advanced. The weather type of 1921 was probably 
the norm instead of the exception. 

Shortly after 1750 this continental type changed to a more 
oceanic type, with milder winters and cooler, rainier summers. 
In England the change seems to have taken place rather 
abruptly in 1752 and was attributed by the commonalty to 
the change of the calendar in that year. This oceanic type 
continued for about a century, culminating in 1850 with 
another maximum advance of the glaciers. There was a 
brief return of the continental type from 1794 to 1810, analysed 
by G. E. P. Brooks (10), which gave a famous period of severe 
winters in Western Europe. 

4. Since 1850 winter temperatures have tended to rise 
over all the north temperate and Arctic regions and probably 
in corresponding latitudes of the Southern Hemisphere.* 
The change was slow and irregular at first, but became very 
rapid after 1900. The rise in the mean temperature of the 
three winter months, from 1851-1900 to 1901-1930, amounted 
to 5 F. or more in Western and Central Europe. This 
change was associated with a marked strengthening of the 
atmospheric circulation and steady west-south-west winds in 
Western Europe. There was little change of summer tem- 
perature. Glaciers and ice-sheets receded very rapidly, and 
after 1918 little or no drift ice reached the shores of Iceland. 
The rise of winter temperature progressed from south to 
north, and Central Europe may have passed the crest as early 
as 1920 when the rise in the Arctic was in full swing. The 
magnitude of the change in the Arctic is shown by the mean 
winter temperatures of Spitsbergen, which rose by 16 F. 
between 1911-1920 and 1931-1935. The edge of the main area 
of Arctic ice also receded towards the pole by some hundreds 
of miles. Since January 1940 the winter climate of Europe 
has reverted abruptly to greater severity, but it is too soon 
to say whether this is the beginning of another long period of 
continental climate or only a temporary fluctuation. 

This concludes the examination of historical changes of 
climate, and also the analysis of the causes of climatic variations. 
The problem has proved to be one of great complexity, but 
throughout the book I have tried to examine each suggested 


cause impartially. The results seem to me to point very 
strongly to the following conclusions : 

1 . The major climatic oscillations, lasting millions of years, 
are due to the major cycles of mountain-building and 
degradation, and their geographical effects in the widest sense, 
which possibly include variations in the amount of carbon 
dioxide and volcanic dust in the atmosphere. 

2. Climatic oscillations of the second order, lasting 
thousands or tens of thousands of years, are due to two or 
possibly three causes : 

(a) Minor changes in the land and sea distribution, caused 
partly by the shifting of the load on the earth's crust by 
erosion and partly by the isostatic effects of the growth 
and decay of the ice-sheets themselves. These were 
mainly effective during periods of high orography. 

(b) Astronomical changes eccentricity of the earth's 
orbit, obliquity of the ecliptic, precession of the 
equinoxes and possibly other causes. These are con- 
tinuously effective and can be traced in some of the 
warm periods. They may have caused the succession 
of glacial and interglacial periods. 

(c) Possiblylong period variations of solar activity. Clima- 
tic oscillations of a few hundred years appear to be 
related to solar changes and it is a reasonable inference 
that the range of solar activity in the course of tens of 
thousands of years has been greater than the range 
during the Christian era. If such changes did occur, 
they must have caused considerable changes of 

3. Climatic oscillations lasting a few hundred years. So 
far as the evidence goes, these seem to be due mainly to 
variations of solar activity. 

4. Climatic oscillations lasting for shorter periods, up to a 
hundred years or so. These may be due in part to variations 
of solar activity but there is evidence that they are often due 
to changes in the general circulation of the atmosphere which 
may have no external cause. They result from the inter- 
action ($ the winds, ocean currents and floating ice-fields 
which we know to occur, but which, in the present state of 
our knowledge, is incalculable. These changes must always 


have occurred, but were probably on a smaller scale during 
the warm periods, when there were no polar ice-caps, than 
during the glacial periods. 

Other factors, hitherto unsuspected, may be discovered and 
prove to be important, but as a result of this analysis there 
seems to be no necessity to introduce hypothetical agents such 
as clouds of cosmic dust, strange stars or great disturbances 
of the earth's axis of rotation. The known causes set out in 
this book suffice to account for the variations of climate during 
geological and historical time. 


(1) WALKER, SIR GILBERT. " Correlations in seasonal variation of weather. 

VIII. A preliminary study of world weather." Calcutta, Indian Met. 
Mem., 24, Pt. 4, 1923. * 

(2) . idem. " VI. Sunspots and pressure.'' idem., 21, Pt. 12, 1915. 

(3) HUNTINGTON, E., and S. S. VISHER. " Climatic changes, their nature and 

causes." New Haven, 1922. 

(4) BROOKS, C. E. P. " The variation of the annual frequency of thunderstorms 

in relation to sunspots." London, Q.J.R. Meteor. Soc., 60, 1934, P- J 53- 

(5) Zurich, Vierteljahrschrif*, 38, 1893, p. 77 ; and Washington, D.C., Monthly 

Weather Rev., 30, 1902, p. 173. 

(6) TURNER, H. H. " On a simple method of detecting discontinuities in a 

series of recorded observations, with an application to sunspots." London 
Mon. Not. R. astr. Soc., 74, 1913, p. 82. 

(7) PETTERSSON, O. " Climatic variations in historic and prehistoric time." 

Svtnska Hydrogr.-Biol. Komm. Skriften, 5, 1914. 

(8) BROOKS, C. E. P., and J. GLASSPOOLE. " The drought of 1921." London, 

Q. J. R. Meteor. Soc., 48, 1922, p. 139. 

(9) BROOKS, C. E. P. " The climate of the first half ol the eighteenth century." 

London, Q. J. R. Meteor. Soc., 56, 1930, p. 389. 

'10) BROOKS, C. E. P. " Winds in London during the early i9th Century." 
London, Meteor. Mag., 67, 1932, p. 56. 



The Age of the Earth. Various methods have been suggested 
from time to time by which we can determine the approximate 
period which has elapsed since the formation of the first solid 
crust of the earth ( i ) . The older methods were based on the 
thickness of sedimentary rocks or the amount of salt in the 
ocean, divided by the present rate of accumulation ; they 
assumed that the present rate of geological processes is a fair 
average of their rate throughout geological times. This 
assumption we now know to be false ; the present is a period 
of abnormally high relief, and in addition the unconsolidated 
deposits of the last ice-age facilitate denudation. The 
maximum age of the oldest rocks calculated from the rate 
of denudation is 350 million (3'5Xio 8 ) years, and this is 
probably only about one-fourth of their real age. 

Presumably the sun is older than the earth, and calculations 
of the age of the sun based on the supply of energy by the 
aggregation of hydrogen atoms gives 140,000 million (i*4X 
io ix ) years as the age of the sun. The sun apparently existed 
alone for a long period before a passing star disrupted it to 
form the solar system, for a calculation of the time since 
Mercury first took shape as a planet indicates that the age of the 
solar system is probably not greater than 10,000 million 
(io 10 ) years. 

The most reliable method of calculating the age of any 
particular portion of the earth's crust is based on the 
phenomena of radio-activity. As is now well known, the 
elements uranium and thorium are continually breaking up 
and passing through a series of changes, the end-products of 
which are lead and helium. So far as we know, in the natural 
state the rate at which each of these elements disintegrates 
is a peculiarity of the element itself, and is entirely independent 
of the physical changes, such as variations of pressure and 
temperature, which it undergoes. If a sample of rock contains 
a certain amount of uranium, at the end of about 5,000 



million years it will contain half this amount, at the end of 
another 5,000 million years one-quarter, and so on, the amount 
remaining continually halving in 5,000 million years. The 
original mass of uranium will never quite disappear. Suppose, 
now, that a rock when it solidified contained a certain amount 
of uranium, but no lead or helium. To-day it contains 
uranium, lead, and helium, and from the ratio of the amount 
of uranium to the amount of lead, or to the amount of helium, 
we can calculate the number of millions of years which have 
elapsed since that particular rock was formed. The uranium- 
lead ratio gives the more reliable estimates, for helium, being 
a gas, is liable to escape, and the uranium-helium estimates 
are systematiceilly too low. In this way the following ages 
(in millions of years) have been calculated for different 
geological periods : 

Oligocene ... 26 

Eocene .... 60 

Carboniferous . . 260-300 

Devonian . . 310-340 
Archaean . . . 560-1,340 

In 1947, A. Holmes (2) concluded that " on the evidence 
at present available, the most probable age of the earth is 
about 3,350 million years." 

J. Joly (3) developed an interpretation of the geological 
history of the earth, which leads him to much smaller values 
for the ages of the rocks. Owing to the universal presence 
of radio-active material in the earth's crust, both sial and 
sima, there is a perpetual generation of heat. In the conti- 
nental masses of sial this heat is able to escape, but in the 
deeper layer of sima it is unable to escape and so goes on 
accumulating until the sima reaches its melting point. Melting 
begins at a considerable depth and proceeds gradually upwards. 
Finally, melting extends to such a height that the accumulated 
heat is able to escape into the oceans, and the substratum 
again becomes solid. Owing to the changes of density and 
volume involved, the period of melting is one of continental 
subsidence, while the period of solidification is a time of 
mountain-building. Melting escape of heat solidification 
constitute a cycle, and according to Joly's calculations a 
cycle requires from forty to sixty million years to consummate 
itself. The number of complete cycles recognised is quite 
small four or five, according to different authors. The 


periods of mountain-building closing the cycles occur as 
follows :- Laurentian and Algoman revolutions in the 
Archaean, Huronian or Killarney closing the Archaean, Appa- 
lachian or Hercynian closing the Palaeozoic, and Alpine in the 
Miocene or Pliocene. Other revolutions which are recognised 
by some geologists but not by others are the Caledonian, 
occurring in the Silurian, and the Laramide in the Cretaceous. 
Hence the total duration of time since the Laurentian can have 
been only 200 to 300 million years. This estimate is quite 
incompatible with the usually accepted data of radio-activity 
given by the uranium-lead ratio, and Joly seeks to explain 
the discrepancy by supposing that the speed of radio-active 
processes has not in fact been constant during geological time, 
but that part of the lead contained in the early rocks was 
formed from isotopes of uranium which disintegrated at a 
greater rate than the only form known at present. There are 
minute peculiarities in some of uranium effects in the early 
rocks which support this view, but it does not seem probable, 
since the longer periods fit in better with the mass of geophysical 
data. A. Holmes (4), reviewing Joly's work, states that the 
generally accepted ages of the rocks are not likely to be in 
error by more than 10 per cent. He accepts the period of 
about forty million years for one of the cycles, but thinks that 
there have been about five times as many cycles as Joly 
supposes, and suggests that the main revolutions recognised 
by Joly arc the concluding stages of major cycles in which 
the melting and consolidation of the deeper seated magma 
is added to that of the more superficial magma which by 
itself formed only minor revolutions. Inspection of the later 
part of the geological record seems to support Holmes' view ; 
for instance, minor periods of orogenesis occurred at the end 
of the Jurassic, in the Upper Cretaceous, in the Oligocene- 
Miocene, and in the Quaternary. The age of the Eocene is 
given by the uranium-lead ratio as 60 million years, so that 
we may take 80 million years for the interval between the 
Upper Cretaceous mountain-building and the Quaternary, 
or a period of two minor cycles. 

If we take the total thickness of the stratified rocks and 
calculate the time which would be required for their formation 
on the assumption that denudation has always proceeded at 
its present rate, we obtain for the age of the oldest rocks only 











| s 


. bo 

. . 

bo S 


^ i-S 



^C S > 



; p q 

e cS 



cr cr 



I | %> 

8 ^' 
J5 2 8 


CS .t 

^ iJ 
o3 cj c3 

. % % % 


03 2 




_; u o u 

2 a .s S 
8 a g a - 


TO > < 





^"S ^2 






bo ^ "+j 
<! C UN ^ 

; - 


o o o 

- 1 iO CD 








O T^ O -" 
- co -^ 10 


" ^^ c^ O 




- -2 6 g 


__. N 

bo w 2 "^ x 


- co ^ r>. 

CO iO ^ 



10 1-1 co r^ 



^WQ ^ 2 



CO CO ^3"* "^h 1 


, e 

a c 8 


-. d " d 

j- coco 

iO d CD 



co co d ^t 
co "* r^ cO 


* ' 

Q ~*> 

A i o'rt 

1.0 >-. co 

CD r^ co 







*-i 04 



o ^ - o 

d d H 


c . }j 

g > d **" . 


CO d CO 



^00 CO 


CO d (N d 

d O 



- CO O d 


i VM ^3 tS 


co xf d o 





d - HH d 










o 2-g 































about 150 million years, which is much too low. A much 
better agreement is obtained if we assume that in each period 
the rate of denudation has been proportional to the average 
elevation. If for any geological period we divide the total 
thickness of the stratified rocks by the average elevation during 
that period (Table 31), we get a number which we may 
term the " duration-ratio " of that period. The sum of all 
the duration-ratios from the beginning of the Keweenawan 
series (Upper Proterozoic) to the close of the Pliocene is 
approximately 156. The age of some probably Upper 
Proterozoic uraninite from Morogoro, East Africa, is given by 
the uranium-lead ratio as 560 million years. Hence a 
" duration-ratio " of unity corresponds with an actual duration 
of 3-6 million years. Similarly, the sum of the duration- 
ratios since the beginning of the Devonian is 107, while the 
greatest age determined from the uranium-lead ratio for a 
Devonian rock is 340 million years, giving a value of 3-2 
million years for a duration-ratio of unity. The corresponding 
value deduced from the Carboniferous radio-active rocks 
(duration-ratio 87, age 300 million years) is 3-4, and from the 
Eocene (duration-ratio 18, age 60 million years) is 3-3. 
These four values are in sufficiently good agreement, and we 
can adopt as the time-equivalent of a duration-ratio of unity 
the mean period of 3 4 million years. This gives us the ages 
and durations of the various geological periods shown in Table 
31, which are in sufficiently good agreement with the 
determinations by radio-active ratios. 


("i) JEFFREYS, H. " The earth, its origin, history, and physical constitution." 
2 ed. Cambridge, 1929. 

(2) HOLMES, A. " A revised estimate of the age of the earth." London, Nature, 

'59> 1947, P- 127. 

(3) JoxYj J- " The surface-history of the earth." Oxford, 1925. 

(4) HOLMES, A. " Radio-activity and geology." Nature, 116, 1925, p. 891. 



The various theories of the causes of climatic change are set out 
below, classified into types. The intention is not to provide 
a complete bibliography, but to give some indication as to 
where a description of the theory can be most readily found. 


CALVERWELL, E. P. Geol. Mag., 32, 1895, p. 64. [Gas-filled 

regions in space.] 
GUILLEMIN, . Arch. sci. phys., (3) 22, p. 585. [Cosmical 

dust in space.] 

HOYLE, F., and R. A. LYTTLETON. See Chapter IV. 
IVES, R. See Chapter IV. 
NOELKE, F. " Das Problem der Entwicklungsgeschichte 

unseres Plane tensystems." Berlin, 1908. [Cosmical dust 

in space.] 

DUBOIS, E. See Chapter IV. 

FAYE, . " Concordance des epoques geologiques avec les 
epoques cosmogoniques." Paris, C. R. Acad. 6a., 100, 
1885, P- 9 26 - 

FISCHER, E. " Eiszeittheorie." Heidelberg, 1902. [The sun 
moves in an elliptical orbit ; when far from focus velocity 
falls off and temperature decreases, giving an ice-age.] 

HUNTINGTON, E., and S. S. VISHER. See Chapter IV. 

JAEKEL, O. %s. D. Geol. Ges., 57, 1905, Monatsber., p. 223 
[The separation of each of the inner planets from the sun 
was accompanied by a decrease of radiation and an ice-age 
on the earth ; obsolete.] 



ADHEMAR, J. " Les revolutions de la mer. Deluges perio- 
diques." Paris, 1842. [Glaciation with winter in 
aphelion at maximum eccentricity of earth's orbit.] 

BALL, Sir R. "The cause of an ice-age." London, 1891. 
[Precession of the equinoxes.] 



CROLL, J. " Climate and time in their geological relations." 

London, 1875. [Eccentricity, glaciation with winter in 

EKHOLM, N. London, Q. J. R. Meteor. Soc., 27, 1901, p. 27. 

[Obliquity of ecliptic.] 
HILDEBRANDT, M. " Die Eiszeiten der Erde." Berlin, 1901. 

[Glaciation with small eccentricity.] 
MILANKOVITCH, M. See Chapter V. 
MURPHY, J. J. London, Q,. J. Geol. Soc., 32, 1876, p. 400. 

[Glaciation with summer in aphelion at maximum 


PETTERSSON, O. (Tidal variations.) See Chapter XXII. 
SPITALER, R. See Chapter V. 


HOFFMANN, J. F. Beitr. Geophys., 9, 1908, p. 405. [Warm 
periods due to heat set free by decomposition of organisms 
in strata.] 


PROBST, J. " Klima und Gestaltung der Erde in ihren 
Wechselwirkungen." Stuttgart, 1887. [Warm springs.] 

WAGNER, A. (Radio-active heat.) See Chapter X. 

We may include here : 

FRANKLIN, A. V. Toronto, J. R. Astr. Soc. Canada, 12, 1918, 
p. 450. [Radiation from a warm moon.] 


KREIGHGAUER, D. " Die Aquatorfrage in der Geologic.'* 

Steyl, 1902. [Polar movements.] 

SIMROTH, H. " Die Pendulationstheorie." Leipzig, 1907. 
OLDHAM, R. D. Geol. Mag., 23, 1886, p. 300. 
WEGENER, A. Sec Chapter XIII. 


ENQUIST, F. Bull. Geol. Inst. Upsala, 12, 1915, p. 35. [Deep- 
ening of ocean basins.] 

GREGOIRE, A. Bull. Soc. Beige Geol., 23, 1909, p. 154. [The 
sea floor is colder than the land, hence when there is a 
reversal the sea is warmed, increasing evaporation, which 
causes greater snowfall on cold land.] 

LE CONTE, J. " The Ozarkian and its significance." J. Geol. y 
Chicago, 7, 1899, P- 5.25- 

RAMSAY, W. Ofversigt afFinska Vetenskaps Soc. Forh., 52, 1910, 
Afd. H. See also Chapter X. 

SCHUCHERT, CH. Carnegie Inst., Washington, Publ. 192, 
.,1914, p. 263. 

UPHAM, W. Amer. Geol. y 6, 1890, p. 327. 




BROOKS, C. E. P. See Chapter VIII. 

HARMER, the late F. W. London, Q,. J. R. Meteor. Soc., 51, 

!925> P- 247. 
KERNER, F. v. Wien, Sitzungsber. K. Akad. Wiss., Math.-nat. 

Kl., 122, Abt. 20, 1913, p. 233. 
LYELL, CH. "Principles of geology." nth ed. London, 

SEMPER, M. s. D. Geol. Ges., 48, 1896, p. 261. 


CHAMBERLIN, T. C. See Chapter III. [Reversal of deep sea 

HULL, E. London, Q. J. Geol. Soc., 53, 1897, p. 107. 

[Deflection of Gulf Stream by Antillean continent.] 
KLEIN, H. J. Gaea, 41, 1905, p. 449. [Deflection of Gulf 

Stream by land projection from Newfoundland towards 

Cape Verde Islands.] 


ARRHENIUS, S. Phil. Mag., 41, 1896, p. 237. [Carbon 


CALLENDAR, G. S. (Carbon dioxide.) See Chapter VI. 
CHAMBERLIN, T. C. J. Geol., Chicago, 7, 1899, pp. 545, 667, 

752. [Carbon dioxide.] 
FREGH, F. ^s. Ges. Erdk., Berlin, 37, 1902, p. 611. [Carbon 

HARBOE, E. G. %js. D. Geol. Ges., 50, 1898, p. 441. [Water 

vapour from volcanoes.] 
HARL, E. and A. Bull. Soc. Geol. France, n, 1911, p. 118. 

[Probable greater pressure of air in geological times. See 

Chapter II.] 


HUMPHREYS, W. H. See Chapter VI. 

SARASIN, P. and F. Basle, Verh. Naturf. Ges., 13, 1901, p. 603. 


ABBE, C. Washington, U.S. Weather Bureau. Monthly 

Weather Review, 34, 1906, p. 559. [Slight changes of 

DEELEY, R. M. Geol. Mag., (6) 2, 1915, p. 450. [Effect of 

great polar water areas on stratosphere.] 
DINES, W. H. See Chapter II. 
HARMER, F. W. See Chapter II. 


Abbe, C., 386 

Abbot, C. G., ioj, 1 1 8, 120 

Abert Lake, 350 

Abyssinia, rainfall, 330, 338 

Adhemar, J., 384 


Climate in historical period, 


pluvial periods, 276 

post-pluvial, 339 
Aftonian, 242 

glaciation, 242 
Tertiary climate, 242 
Aldrich, L. B., 112, 1 20 
Alexandria, climate in first 

century, 333 

Algoman revolution, 381 

climate, 226 

mountain-building, 205 

folding, 178 

revolution, 381 
Alps, traffic over, 143, 301 
America, climate in historical 

period, 3420. 

Amorite migration, 292, 320 
Anau, 318 

Angaraland, 247, 248, 257 
Angkor, ruins of, 326 


Angstrom, A., 112, 120, 123, 126, 


Carboniferous climate, 259 

climate in geological times, 243 

glaciation of, 239 

pressure and winds, 64 
Antevs, E., 1 13, 120, 181, 191, 236, 
243, 246, 263, 273, 277, 
278, 342, 358 

continental winter, 46, 60 

glacial, 641!. 

sub-tropical, 46, 55, 61 
Antillean Continent, 8 1 , 386 
Antiochus, Calendar of, 335 

Aphelion, 102, 364 
Appalachian revolution, 38 1 
Arabian migration, 292 
Arago, 286, 294 
Aral, Sea of, 322 
Aramaean migration, 292, 320 
Arch<ocyathin<e, 23, 226, 245 
Arctic Ocean, ice in, 416., 1391!., 

268, 371, 376. 
Arctowski, H., 119, 121 
Arizona, archaeology, 353 
Arkell, W. J., 194, 197 
Arldt, T., 203, 205, 220, 247, 262 
Arrhenius, S., 386 
Aryans, 291, 320 
Asama, Mt., 1 18 
Asia, climate in historical period, 

3 1 8ft"., 359 

Astronomical theories, iO2ff., 384 
Atlantic period, 142, 147, 173, 

296^-, 339 
Atlantis, 248, 257 
Atmosphere, composition of, 

ii3lT., 386 

circulation, 49!!. 

changes of, 273, 373, 386 
zonal, 54 

Aurora and sunspots, 366 
Australia, climatic history, 244 

Bacon, Sir F., 313, 317 

Ball, Sir R., 384 

Baltic stadium, 106 

Beadnell, H. J. L., 335, 341 

Belgium, Meteorological Chron- 
icle of, 287 

Benguela Currents, 76, 132 

Berg, L., 322, 327 

Berry, E. W., 24, 27, 228, 241, 

" Big Trees," see Sequoias. 

Bishop, C. W., 324, 328 

Bjerknes, J., 67 

Boden See, variations of level, 






climate, 173 
period, 2g6fF., 364 
zone (faunal), 23 
Bradley, W. H., 108, 109, 310 
Brahe, Tycho, 284, 294, 312 
Britton, C. E., 289, 294, 303, 304, 


Bronze Age, 296, 298 
Brooks, C. E. P., 33, 45, 67, 99, 
101, 119, 121, 157, 164, 
i?5 312, 317, 3 6 5> 375- 
376, 378 
Bruckner, E., 93, 101, 154, 290, 

294> 32i, 327 
Bryan, K., 264, 273, 277, 278, 

35 1 > 358 

" Burial of Olympia," see Hunt- 
ington, E. 


folding, 178 

revolution, 381 
California Current, 132 
Callendar, G. S. } 117, 120 
Calverwell, E. P., 384 
Gandee, H. C., 328 
Carbon dioxide, 1 16, 265, 386 

climate of, 108, 247fF. 

deserts, 237 

geography, 248 

gl-aciation, see Permot-arboni- 

mountain-building, 178, 205, 


ocean currents, 248, 250 
volcanic action, 211, 250 
winds, 252 
Carruthers, N., 309 
Caspian, variations of rainfall, 321 
Caton-Thompson, G., 335, 341 
Ceratodus, 26, 172, 194, 226 
Chamberlin, T. C., 79, 88, 116, 

203, 220, 386 

Chandler, M. E. J., 96, 101 
Chancy, R. W., 196, 241, 246 
Chewaucan Lake, 350 
Childe, V. G., 316 
Chile, variations of rainfall, 358 
China, climate in historical period, 

322, 3.25 
Chinese archives, 325, 366 

geological, 379 
of ice age, 106, 107 
post-glacial, 296 
Chu, Co-Ching, 325, 328 
Gimbrian flood, 302 

atmospheric, see Atmospheric 


oceanic, see Oceanic circulation 
" Civilisation and climate," see 

Huntington, E. 

change, theories of (listed), 384 
factors, astronomical, loiff. 
geographical, 20 iff. 
solar, 89ff. 
Optimum, 105, 142, 297, 362, 

zones, 23, 174 
Close, Sir C., 222, 230 
Cloudiness, I22if. 

and elevation, 183 

and temperature, I22ff., 183 
Clouds, reflection of radiation by, 

I 12 

Coal Measures, 171, 236, 257 
Coleman, A. P., 231, 233, 245, 246 
Golima, eruption of, 113 
Constance, Lake, 3OofT. 
Continental drift, 22 iff. 

and glaciation, 154 

and temperature, I29ff., 203, 

variations of, 108, 204 
Coral reefs, 194, 225 
Gordilleran ice-sheets, 273 
Correlation coefficients, 210, 214 

dust, 384 

theories of climatic change, 384 
Cretaceous, 108 

climatic zones, 23, 239 

deserts, 25 

geography, 240 

glaciation, 177, 216, 228 

ocean currents, 240 
Croll,J., 103, 109, 385 
Crowther, E. M., 168, 176 
Cycles, see Periodicities 
Cyclone tracks, 98, 160, 273 
Cyrenaica, climate ir/ historical 
period, 333 



Dacque, E., 108, 109, 205, 220 
Ball, W. H., 242, 246 
Date palm in Palestine, 286 
David, Sir T. W. E., 233, 235, 244, 

245, 250 

Dead. Sea, level of, 322 
Deeley, R. M., 386 
Defant, A., 119, 121 
de Geer, G., 84, 103, 243 
Deluge, Noachian, 289 
Depressions, see Cyclones 
Deserts, 24, 126, 167, 172, 237, 


Desiccation, 169, 286 
Devonian, 24 
Dines, W. H., 66, 67, noff., 120, 

Doldrums, 49, 174, 276 

Dorians, 320 

Dorsey, H. G., 55, 67, 264, 277 

Drayson, Col., 102 

Drought Commission, South 

Africa, 176 
Droughts, 304!!. 

Dry period, sixteenth century, 373 
Dubois, E., 99, 101 
Duration-ratio, 382 
Durst, C. S., 88 
Dwyka tillite, 235 


age of, 379 

heat, 22, 127, 179, 385 
East Greenland Current, 75, 143 
Easton, C., 310, 317 
Eccentricity of earth's orbit, 102, 

1 08, 384 

Ecliptic, obliquity of, 102, 364 
Ekholm, N., 385 
Elamites, 320 

and glaciation, 1 78 

effect on temperature, 184, 210, 


variations, 203 

England, rainfall records, 307!!. 
Enquist, F., 385 
Eocene, 108 

climate, 96 

flora, 24 

glaciation, 177 

temperature, 139!!. 

Equinoxes, precession of, 103 
Erdtman, G., 296, 297, 316 
Eskimos, 357 

climate in historical period, 

variations of rainfall, 299, 305, 


of temperature, 310, 361 
Eurydesma cot 'datum , 234 
Evaporation, 165, 185, 297 

Equatorial Current, 71, 253 

, -., 384 
Fenno-Scandian moraine, 295 
Ficker, H. v., 276, 278 
Fischer, E., 384 
Fleure, H. J^, 84, 88 
Flint, R. Fl, 55, 67, 264, 274, 277, 


Flood legend, 289 
Floods, 306 

Forbes, J. D., 132, 157 
Fowle, F. E., 118, 120 

A. V., 385 

B, 117 
Freeh, F., 386 

Frost in Miocene, 26, 196 

Gale, H. S., 349, 352 

Gamblian pluvial, 276 

Gams, H., 106, 143, 157, 299, 316 

Gangamopteris, 234 

Gardner, E. W., 335, 34 1 

Gas-filled regions in space, 384 

Geer, G. de, see de Geer, G. 

Geikie,J., 154 

Geographical factors of climate, 

20 iff., 386 

Geological periods, ages of, 382 

anticyclones, 64 

deposits, correlation of, 228, 
242, 264 

periods, see Ice ages 
Glaciers, expansion of, historical, 

Glasspoole,J., 308, 378. 

Glossopteris, 234, 236 
Godwin, H., 298, 316 
Gondwanaland, 247ff. 
Granlund, E., 298, 316 
Grant, C. P., 319, 327 
Greece, variations of climaiu, 31^ 




climate in historical period, 143, 


inland ice, 52 
weather, 63 
Gregoire, A., 385 
Gregory, J. W., 322, 327 
" Grenzhorizont," 298 
Guiana Current, 81 
Guillemin, -., 384 
Gulf Stream, 66, 72, 81, i34 ff - 3 86 
Gumban culture, 339 
Gunz glaciation, 84, 93, 106, 107, 

242, 269 

Gunz-Mindel interglacial, 93, 242 
Gypsum, deposits of, 25, 172, 237, 


Haddon, A. C,, 327 
Hail in Tertiary, 163 
Hallstatt period, 300 
Handlirsch, A., 250, 262 
Hansen, H. P., 350, 358 
Harboe, E. G., 386 
Hardy, E. M., 298, 316 
Harle, E. and A., 120, 121, 386 
Harmer, Sir F., 66, 67, 84, 88, 

258, 262, 386 
Hasluck, M., 302, 316 
Haurwitz, B., 123, 128, 255 
Hedley, G., 153, 157 
Heer, O., 136, 157 
Helium, 379 

Hellmann, G., 284, 294, 333, 341 
Hennig, R., 317 
Hepworth, M. W. G., 88 

folding, 178, 233 

revolution, 381 

Herjolfsnes, excavations at, 356 
Herodotus, 322, 331, 340 
Higgins, L. S., 311, 31? 
Hildebrandsson, H. H., 285, 294 
Hildebrandt, M., 385 
Himpel, K., 96, 101 
Historical period, climates of, 

Hobbs, W. H., 63, 67, 163, 175 
Hobley, C. W., 340, 341 
Hoffmann, J. F., 385 
Hohokam, 354 
Holland, rainfall, 309 
Holmes, A., 380, 383 
Housa State, 338 

Hovgaard, \V., 356, 

Hoyle, F., 94, 101 

Hull, E., 386 

Humboldt Current, 76, 132, 271 

Humphreys, W. J., 117, 120 

Hunt, T. M., 119, 121, 164, 175, 

3 I2 > 3 1 ? 
Huntington, E., 

" Burial of Olympia," 294, 315, 

3 1 ?* 322 
" Civilisation and climate," 

292, 294, 356, 358 
" Climatic changes, their nature 

and cause," 86, 88, 98, 

" Climatic factor," 342, 353, 

" Mainsprings of civilisation," 

319, 321, 327 
4< Pulse of Asia," 321, 327 
" Pulse of progress," 314, 317 
" World power and evolution," 


Huronian revolution, 381 
Hyksos, 291, 320 

Ice ages, 

causes of, 265 

astronomical, 103(1'. 
carbon dioxide, 116 
interstellar matter, 94, 384 
mountain-building, 178 
Nova outbreaks, 96 
ocean currents, 84 
sea temperature, 266 
solar radiation, 91 
sunspots, 98 
Permo-Carboniferous, icu, 

23iff., 257 
Pre-Cambrian, 180 
Quaternary, 177, 215, 263!!'. 
Ice, cooling effect of, 32rT., 184 
Iceland, climate in historical 

period, 303 

area of, 113, 263 
nourishment of, 163 
weather over, 63 
winds over, 62 
Ihering, H. v., 196, 197 
Illinoian glaciation, 242 
India, climate in historical period, 

3*4 . f 

Insects, holometabolism, 250 


39 * 

Instrumental records of weather, 


Interglacial periods, 94, 242, 273 
Interstellar matter, 94, 384 
Ireland, climate in historical 

g period, 302 
Islarmtic expansion, 292 
Isoflors, Eocene, 241 
Istakhri, 322 
Ives, R. L., 100, 10 1 

Jaekel, O, 384 

Jafnites, 320 

Jana Sea, 138 

Jeffreys, H., 53, 67, 179, 191, 204, 

220, 224, 230, 383 
Jerseyan glaciation, 242 
Joly,J., 380, 383 
Jones, J. C., 346, 358 
Jwlien, M., 233 

climate, 194 

deserts, 25 

faunal zones, 23 

temperature, 141 

Kafuan, pluvial, 276 

Kamasia, Lake, 95, 276 

Kansan glaciation, 242 

Kashmir, glaciation, 265 

Kassites, 320 

Katmai, eruption of, 113, 118 

Keane, A. H., 341 

Keewatin ice-sheet, 273 

Kenriard, A. S., 299 

Kerner, F., 41, 45, 80, 95, 101, 

134, i37fF., 157, 176, 386 
Kessler, P., 274, 278 
Khanikof, -., 32:; 
Kharga oasis, 335 
i&llarney revolution, 381 
Klein, H.J., 386 
Koch, L., 144, 157 
Koppen, W., 104, 109, 169, 176, 

225, 230, 231 

Krakatoa, eruption of, 113, 118 
Kreichgauer, D., 385 
Krige, L.J., 128 
Kubierschky, -., 25, 195 

Labrador Current, 74ff., 137, 268 
Labradorean ice-sheet, 274 

Labrijn, /\., 311, 317 

La Cour, P., 284, 294, 312, 


Lagrange, J. L., 102 
Lahontan, Lake, 346ff. 
Lake-dwellings, 299 
Lakes, fluctuations of, 276, 290, 

297> 346ff. 
Lang, R., 167, 176 
Laramide revolution, 381 
Lasareff, P., 78, 88, 208, 251 
La Tene stage, 301 
Laterite, 167 

Laurentian revolution, 381 
Leaching factor, soils, 168 
Leakey, L. S. B., 341 
Le Conte, J., 180, 191, 385 
Leiter, H., 333, 340 
Leverett, F., 242, 246 
Lias, 216 

period, i47fF. 

submergence, 362 
" Little Ice-Age," 301, 375 
Loess, 275 
Loewc, F., 265, 277 
Lop Nor, 323 
Lowe, E. j., 304, 317 
Lung-fish, see Ceratodus 
Lyell, C., 386 
Lyons, Sir H., 330, 340 
Lyttleton, R. A., 94, 101 

Mackay, E. J. H., 324, 327 
Makalian pluvial, 277 
Malaria, 314, 355 
Mandingaii Empire, 338 
Mauley, G., 180, 191, 311, 317 
Manson, M., 126, 128, 161, 175 
Mariolopoulos, E. G., 316, 317 
Markham, S. F., 293, 294 
Mar tonne, E. de, 191 
Matthes, F. E., 63, 67, 301, 316 
Mayan civilisation, 355 
" Mediterranean " climate, 59, 


Meinardus, W., 64, 67 
Merle, W., 284, 294, 312 
Mesozoic, 120, 138, i92ff., 227 

deserts, 24, 238 
Meteorites, 1 19 

Meyer, G. M., 303, 312, 316, 317 
Mexico, Lake of, 353 
Migrations, 291, 320 

from China, 326 
Milan rainfall, 309 
Milankovitch, M., 102, 104, 109 


Mindel glaciation, 93, 106, 107, 

Mindel-Riss interglacial, 94, 242, 

264, 270, 273 

frost, 196 

see also Tertiary 
Mohenjo-Daru, 324, 360 
Mono, Lake, 346 
Monsoon rainfall, 171 
Monsoons, 61, 254 
Moreau, R. E., 277, 278 
Mortensen, H., 271, 277 
Moslem civilisation, 315 
Mossi State, 338 
Mountain-building, 205, 381 

and climate, 128, 177*?., 385 
Miihlbergian glaciation, 106 
Murphy, J. J., 104, 109, 385 
Murray, G. W., 337 

Nakuran pluvial, 339 
Nearctis, 248, 257 
Nebraskan ice-sheet, 242 
Negri, C., 333, 341 
Neolithic, 296 
Neumayr, M,, 23, 27 
New Mexico, archaeology, 353 

hydrography of, 329 

variations of level, 329ff., 368 
Nilsson, E., 95, 101, 276, 278, 339, 


Noelke, F., 384 
" Non-glacial " 

circulation, 53 

periods, 321!., i39ff. 

temperatures, 4 iff. 
Nordhagen, R., 106, 143, 157, 299, 


Norfolkian, 242 
Norse voyages, 144, 303, 357 
" North water," 72, 79 
Nova outbreaks, 96 

Obic Sea, 138 

Obliquity of ecliptic, 102, 108, 364 


currents, 68fl'. 

variation of, 204(1., 386 
depth of, 85, 362 

circulation, 68fT. 
reversal of, 79 

Oceans as regulators, 85, 267 

Oldham, R. D., 385 

Old Red Sandstone, 24, 108, 226 

Oligocenc, 242. See also Tertiary 

Orogenesis, see Mountain-building 

Osborn, H. F., 242, 246 \ 

Ostrovo, Lake, 302 

Owens Lake, 352 

Oxus, change of course, 323 

Ozarkian, 385 

Palceocrystic ice, 39, 75 
Pala'Ogeography, 20 iff., 247 
Palestine, climate in historical 

period, 286 
Pajaritans, 354 
Pangaea, 221 
Paris, rainfall, 309 
Paschinger, V., 271, 277 
Peake, H. J. E., 84, 88, 291, 294, 

Peat-bogs, 173, 296 

in North America, 350 

tropical, 171 
Pelee, eruption of, 113 
Penck, A., 93, 101, 154 
Pendulation theory, 385 
Periglacial climate, 274 
Perihelion, 102 
Periodicities, 97, 108, 310 
Permian, 25, 195, 234 

deserts, ^37 

geography and ocean currents, 

Permo-Carboniferous, 100, 231 ft*., 

247ff., 257 
Petterssen, O., 143, 157, 357, 358. 

368, 378 
Phrygians, 300 
Planetary circulation, 50 
Plato, 169, 316 
Pleistocene, see Quaternary 
Pliocene, see Tertiary 
Pluvial periods, 95, 271, 276 

climate, 173 

east winds, 53, 1 74 

front, 54, 59 

whirls, 52 

Pole, movements of, 229, 244, 385 
Pollen spectra, 296, 350 
Potonie, H., 171, 176 / 
Precession of equinoxes, 103 



Precipitation, 158!?. 

convectional or instability, 162 

cyclonic, 160 

distribution, 164, 169 

in Quaternary, 59, 166, 275 

in warm periods, 185, i93ff. 

le^l of maximum, 1 59 

orographic, 159 

over ice-sheets, 162 

distribution, 46^. 
in warm periods, 62 

systematic variation, 360 
Priestley, R. E., 239, 243, 246, 

I 259, 262 
Probst, J., 385 
Proterozoic, 177, 215, 226 
Ptolemaius, Claudius, 333 
Pueblos, 354 
pyramid Lake, 346, 352 

" Quaternary Climates," 342, 358 

atmospheric circulation, 55 
chronology, 106, 107 
climate, 263!!'. 

outside ice-sheets, 274 
elevation in, 183, 269 
ice-age, 177, 215, 2631! 
ice-sheets, area, 263 

correlation, 242, 264 

retreat of, 277, 295 
isotherms, 135, 154 
land and sea, 152 
precipitation, 59, 166, 275 
snow-line, 264 
winds, 59 


absorption by atmosphere, i loff. 
by volcanic dust, 1 13, 1 1 7, 255 
by water vapour, 1 14 

reflection from clouds, 


from snow, 1 12, 256 
solar, 89 

measurement of, 89 
and temperature, 361 
variations, 90, 96, 384 
terrestrial, i loif. 
Radio-activity, 179, 379 
Rain, see precipitation 
" Rain-faltor," 167 


fluctuation*, 361 

Africa, 329^, 359 

America, 342fT., 359 

Asia, 3i8fT., 359 

Europe, 297, 305, 359 
records, 3076*. 
zones, 164, 174 
Rain-forests, tropical, 170 
" Raininess," 306, 325 
Raised beaches, 362 
Ramsay, Sir A., 233 
Ramsay, W., 180. 191, 385 
Ray, L.L., 277 
Reeds, C. A., 242, 246 
Regression coefficients, 2 1 1 
Reid, C., 154 
Reid, E. M., 96, 101 
RG-VEDA, 324 
Riss glaciation, 94, 106, 107, 242, 

Riss-Wurm interglacial, 94, 242, 

26 4.. . 
Rome, civilisation and climate, 

Rossby, C. G., 54, 67 

Rudistes, 225, 239, 240 
Russell, I. C., 348 
Russia, South, variations of rain- 
fall, 361 

Sabine Island, 222 

Sahara, climatic fluctuations, 338 

Saki, Lake, 301, 367, 372 

St Wilfrid's drought, 289 

Salinity of oceans, 86 

Salt deposits, 25, 172, 234, 237, 


Salton Sea, 348 
Sand ford, K. S., 277, 278 
Sangamon, interglacial, 242 
Sapper, K., 358 
Sarasiii, P. and F., 386 
Sawyer, L. D., 333, 341 
Sayles, R. W., 245 
Scandinavia, variations of sea 

level, 362 

Scanian glaciation, 242 
Scherhag, R., 66, 67 
Schomberg, R. C. F., 324, 327 
Schostakowitsch, W. B., 301, 316 
Schuchert, C., 261, 262, 385 
Schulman, E., 354, 358 

394 INDEX 


lowering by glacir.tion, 181, 263 

variations of, 361, 373 
Sea temperature and glaciation, 


Seistan Lake, 323 
Semite migration, 320 
Semper, M., 386 
Sequoias, 293, 3421!., 361 
Shaw, Sir N., 51, 67 
Sial, 222, 380 

Siberia, glaciation, 264, 265 
Sicilian beach, 85 
Silurian, 178, 205 
Sima, 222, 380 
Simpson, Sir G., 64, 67, 91, 101, 

114, 163, 175, 262 
Simroth, H., 385 
Smith, E. H., 277 
Smithsonian Institution, 89, 101 

level of maximum, 159, 271 

over ice-sheets, 163, 273 
Snow-line, 159, 264 
Soil colour and climate, 167 
Solar-cyclonic hypothesis, 98 

activity and rainfall, 366 
variations of, 361 
see also Sunspots 

constant, 90 

radiation, see Radiation, solar 
Solberg, H., 67 

Speerschneider, C. I. H., 303, 316 
Spitaler, R., 86, 88, 103, 109, 132, 

I56> 157 

Squantum tillite, 232 
Stein, Sir A., 322 
Stockwell, J. N., 102, 108 
" Storm floods," 303, 372 
Stormmess and sunspots, 98, 365 
Storms, 306 

Storm tracks, 98, 160, 273, 369 
Strabo, 322 
Stratosphere, 51, 386 
Sub- Atlantic period, 173, 296ff., 


Sub-boreal period, 296ff., 340 
Sumerians, 320 
Summer Lake, 350, 352 
Sun, age of, 379 

cycle, cause of, 367 
length, 97, 108 

Sunspots, 97, 366 
and meteors, 367 
and Nile levels, 368 
and rainfall, 98, 365 
and storminess, 98, 365 
and temperature, 98 i 
and thunderstorms, 98, 365 
relative number, 97 
variations of, 97, 366 

Supan, A., 174 

Susa, 319 

Siissmilch, C. A., 233, 244, 245, 

Sverdrup, H. U., 88 

Symons, G. J., 304, 307, 31'; 

Syria, caravan travel, 319 

TarT,J. A., 231, 245 
Tanganyika Lake, 340 
Tansley, A. G., 298, 316 
Taulis, E., 358 

and continentality, I29ff., 203, 

decrease with height, 211, 212 

distribution, 1 29!!. 

over ice-sheets, 33fT., 154 

variations, 2041!. 
*' Terra Nova,*' 246, 262 

climate, 26, 83, 108, 163, 173, 
igafT., 228, 237, 242 

flora, 24, 136, 241 

isotherms, 135 

Tethys Sea, 239, 247, 248, 251 
Theeuws, R., 341 
Thunderstorms and sunspots, 98, 


Tidal friction, 225 
" Tide-generating force," 3^Y 

Tomboro, eruption, 1 1 9 
Torridon Sandstone, 226 
Toussoun, Prince Omar, 329, 340 
Toynbee, A. J. 5 319, 327 
Trade winds, 55, 83 
Quaternary, 83 
Tree growth, 171, 236, 293, 3421!., 


Triassic, 25, 195 
Tripolje, 319 
Tropopause, 51 
Truckee River, 347 



Turner, H. H., 367, 378 

" Twilight of the Gods/' 289 

Uhlig, V., 23, 27, 138 
Umbgrove, J. H. F., 191 
Unafcar, V., 324, 328 
Upilam, W., 385 
Upwelling water, 73, 76, 267 
Uranium-lead ratio, 380 
U.S.A., rainfall of western, 343, 
359, 361 

Vanderlinden, E., 287, 294 

" Varve " clays, 84, 232, 293, 301 

Versregan, R., 312 

Vine in Palestine, 286 

Visher, S. S., 86, 88, 98, 101, 365, 


* action, variations of, 204, 206, 

dust, 113, 117, 255, 260, 270, 

Volga Sea, 248, 251, 258 

Waagen, L., 234 

Wagner, A., 179, 191 

Walker Lake, 352 

Walker, Sir G., 360, 365, 378 

Warming, E., 170, 176 

Warm periods, 

aridity, 24 

atmospheric circulation, 6 1 , 

oceanic circulation, 77^. 

since 1850, 376 

weather of, i92ff. 
Water mills, 303 

Water vapour, effect on climate, 
1 14, 386 

Wayland, E. J., 276, 278 
Weather ^ 

of glacial periods, 274 

of warm periods, 192!! 
Wegener, A., 104, 109, 22 iff., 230, 


Weidmann, S., 231, 245 
White, D., 171, 176 
Willis, Bailey, 261, 262 
Winds, 46ff. 

in historical period, 285, 31 iff. 

in ice-age, 59 

in warm periods, 55 

over ice-sheets, 62 
Winkle, W. van, 350, 358 
Winnemucca Lake, 346, 352 
Winters, severe, 376 
Wisconsin glaciation, 242, 264, 274 
Wittfogel, K. A., 326, 328 
Wolf, R., 366 

Woodworm, J. B., 232, 245 
Wright, C. S., 239, 243, 246, 259, 


Wundt, W., 113, 1 20 
Wurm glaciation, 94, 106, 107, 

242, 265, 270 
Wyville Thomson ridge, 84, 266 

Yarmouth interglacial, 243, 264 
Yucatan, climatic variations, 354 

Zeiller, R., 246 

Zeuner, F. E., 105, 109, 219, 264 

270, 277 
Zimbabwe, 340 

climatic, 174 

of rainfall, 164, 174 

of temperature, 1 74 
warm periods, 23, 241 

wind, 54, 174